stage2: *WIP*: rework ZIR memory layout; overhaul source locations
The memory layout for ZIR instructions is completely reworked. See
zir.zig for those changes. Some new types:
* `zir.Code`: a "finished" set of ZIR instructions. Instead of allocating
each instruction independently, there is now a Tag and 8 bytes of
data available for all ZIR instructions. Small instructions fit
within these 8 bytes; larger ones use 4 bytes for an index into
`extra`. There is also `string_bytes` so that we can have 4 byte
references to strings. `zir.Inst.Tag` describes how to interpret
those 8 bytes of data.
- This is shared by all `Block` scopes.
* `Module.WipZirCode`: represents an in-progress `zir.Code`. In this
structure, the arrays are mutable, and get resized as we add/delete
things. There is extra state to keep track of things. This struct is
stored on the stack. Once it is finished, it produces an immutable
`zir.Code`, which will remain on the heap for the duration of a
function's existence.
- This is shared by all `GenZir` scopes.
* `Sema`: represents in-progress semantic analysis of a `zir.Code`.
This data is stored on the stack and is shared among all `Block`
scopes. It is now the main "self" argument to everything in the file
that was previously named `zir_sema.zig`.
Additionally, I moved some logic that was in `Module` into here.
`Module.Fn` now stores its parameter names inside the `zir.Code`,
instead of inside ZIR instructions. When the TZIR memory layout
reworking time comes, codegen will be able to reference this data
directly instead of duplicating it.
astgen.zig is (so far) almost entirely untouched, but nearly all of it
will need to be reworked to adhere to this new memory layout structure.
I have no benchmarks to report yet, as I am still working through
compile errors and fixing various things that I broke in this branch.
Overhaul of Source Locations:
Previously we used `usize` everywhere to mean byte offset, but sometimes
also mean other stuff. This was error prone and also made us do
unnecessary work, and store unnecessary bytes in memory.
Now there are more types involved into source locations, and more ways
to describe a source location.
* AllErrors.Message: embrace the assumption that files always have less
than 2 << 32 bytes.
* SrcLoc gets more complicated, to model more complicated source
locations.
* Introduce LazySrcLoc, which can model interesting source locations
with very little stored state. Useful for avoiding doing unnecessary
work when no compile errors occur.
Also, previously, we had `src: usize` on every ZIR instruction. This is
no longer the case. Each instruction now determines whether it even cares
about source location, and if so, how that source location is stored.
This requires more careful work inside `Sema`, but it results in fewer
bytes stored on the heap, without compromising accuracy and power of
compile error messages.
Miscellaneous:
* std.zig: string literals have more helpful result values for
reporting errors. There is now a lower level API and a higher level
API.
- side note: I noticed that the string literal logic needs some love.
There is some unnecessarily hacky code there.
* cut & pasted some TZIR logic that was in zir.zig to ir.zig. This
probably broke stuff and needs to get fixed.
* Removed type/Enum.zig, type/Union.zig, and type/Struct.zig. I don't
think this quite how this code will be organized. Need some more
careful planning about how to implement structs, unions, enums. They
need to be independent Decls, just like a top level function.
2021-03-15 23:38:38 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2020-09-13 19:17:58 -07:00
|
|
|
|
2022-03-11 19:31:29 -07:00
|
|
|
|
2020-09-13 19:17:58 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-08-30 19:22:04 -07:00
|
|
|
|
stage2: *WIP*: rework ZIR memory layout; overhaul source locations
The memory layout for ZIR instructions is completely reworked. See
zir.zig for those changes. Some new types:
* `zir.Code`: a "finished" set of ZIR instructions. Instead of allocating
each instruction independently, there is now a Tag and 8 bytes of
data available for all ZIR instructions. Small instructions fit
within these 8 bytes; larger ones use 4 bytes for an index into
`extra`. There is also `string_bytes` so that we can have 4 byte
references to strings. `zir.Inst.Tag` describes how to interpret
those 8 bytes of data.
- This is shared by all `Block` scopes.
* `Module.WipZirCode`: represents an in-progress `zir.Code`. In this
structure, the arrays are mutable, and get resized as we add/delete
things. There is extra state to keep track of things. This struct is
stored on the stack. Once it is finished, it produces an immutable
`zir.Code`, which will remain on the heap for the duration of a
function's existence.
- This is shared by all `GenZir` scopes.
* `Sema`: represents in-progress semantic analysis of a `zir.Code`.
This data is stored on the stack and is shared among all `Block`
scopes. It is now the main "self" argument to everything in the file
that was previously named `zir_sema.zig`.
Additionally, I moved some logic that was in `Module` into here.
`Module.Fn` now stores its parameter names inside the `zir.Code`,
instead of inside ZIR instructions. When the TZIR memory layout
reworking time comes, codegen will be able to reference this data
directly instead of duplicating it.
astgen.zig is (so far) almost entirely untouched, but nearly all of it
will need to be reworked to adhere to this new memory layout structure.
I have no benchmarks to report yet, as I am still working through
compile errors and fixing various things that I broke in this branch.
Overhaul of Source Locations:
Previously we used `usize` everywhere to mean byte offset, but sometimes
also mean other stuff. This was error prone and also made us do
unnecessary work, and store unnecessary bytes in memory.
Now there are more types involved into source locations, and more ways
to describe a source location.
* AllErrors.Message: embrace the assumption that files always have less
than 2 << 32 bytes.
* SrcLoc gets more complicated, to model more complicated source
locations.
* Introduce LazySrcLoc, which can model interesting source locations
with very little stored state. Useful for avoiding doing unnecessary
work when no compile errors occur.
Also, previously, we had `src: usize` on every ZIR instruction. This is
no longer the case. Each instruction now determines whether it even cares
about source location, and if so, how that source location is stored.
This requires more careful work inside `Sema`, but it results in fewer
bytes stored on the heap, without compromising accuracy and power of
compile error messages.
Miscellaneous:
* std.zig: string literals have more helpful result values for
reporting errors. There is now a lower level API and a higher level
API.
- side note: I noticed that the string literal logic needs some love.
There is some unnecessarily hacky code there.
* cut & pasted some TZIR logic that was in zir.zig to ir.zig. This
probably broke stuff and needs to get fixed.
* Removed type/Enum.zig, type/Union.zig, and type/Struct.zig. I don't
think this quite how this code will be organized. Need some more
careful planning about how to implement structs, unions, enums. They
need to be independent Decls, just like a top level function.
2021-03-15 23:38:38 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-02-05 19:39:04 -07:00
|
|
|
|
stage2: *WIP*: rework ZIR memory layout; overhaul source locations
The memory layout for ZIR instructions is completely reworked. See
zir.zig for those changes. Some new types:
* `zir.Code`: a "finished" set of ZIR instructions. Instead of allocating
each instruction independently, there is now a Tag and 8 bytes of
data available for all ZIR instructions. Small instructions fit
within these 8 bytes; larger ones use 4 bytes for an index into
`extra`. There is also `string_bytes` so that we can have 4 byte
references to strings. `zir.Inst.Tag` describes how to interpret
those 8 bytes of data.
- This is shared by all `Block` scopes.
* `Module.WipZirCode`: represents an in-progress `zir.Code`. In this
structure, the arrays are mutable, and get resized as we add/delete
things. There is extra state to keep track of things. This struct is
stored on the stack. Once it is finished, it produces an immutable
`zir.Code`, which will remain on the heap for the duration of a
function's existence.
- This is shared by all `GenZir` scopes.
* `Sema`: represents in-progress semantic analysis of a `zir.Code`.
This data is stored on the stack and is shared among all `Block`
scopes. It is now the main "self" argument to everything in the file
that was previously named `zir_sema.zig`.
Additionally, I moved some logic that was in `Module` into here.
`Module.Fn` now stores its parameter names inside the `zir.Code`,
instead of inside ZIR instructions. When the TZIR memory layout
reworking time comes, codegen will be able to reference this data
directly instead of duplicating it.
astgen.zig is (so far) almost entirely untouched, but nearly all of it
will need to be reworked to adhere to this new memory layout structure.
I have no benchmarks to report yet, as I am still working through
compile errors and fixing various things that I broke in this branch.
Overhaul of Source Locations:
Previously we used `usize` everywhere to mean byte offset, but sometimes
also mean other stuff. This was error prone and also made us do
unnecessary work, and store unnecessary bytes in memory.
Now there are more types involved into source locations, and more ways
to describe a source location.
* AllErrors.Message: embrace the assumption that files always have less
than 2 << 32 bytes.
* SrcLoc gets more complicated, to model more complicated source
locations.
* Introduce LazySrcLoc, which can model interesting source locations
with very little stored state. Useful for avoiding doing unnecessary
work when no compile errors occur.
Also, previously, we had `src: usize` on every ZIR instruction. This is
no longer the case. Each instruction now determines whether it even cares
about source location, and if so, how that source location is stored.
This requires more careful work inside `Sema`, but it results in fewer
bytes stored on the heap, without compromising accuracy and power of
compile error messages.
Miscellaneous:
* std.zig: string literals have more helpful result values for
reporting errors. There is now a lower level API and a higher level
API.
- side note: I noticed that the string literal logic needs some love.
There is some unnecessarily hacky code there.
* cut & pasted some TZIR logic that was in zir.zig to ir.zig. This
probably broke stuff and needs to get fixed.
* Removed type/Enum.zig, type/Union.zig, and type/Struct.zig. I don't
think this quite how this code will be organized. Need some more
careful planning about how to implement structs, unions, enums. They
need to be independent Decls, just like a top level function.
2021-03-15 23:38:38 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2020-09-13 19:17:58 -07:00
|
|
|
|
|
|
|
|
|
2021-07-08 20:42:47 -07:00
|
|
|
|
2021-04-13 12:38:06 -07:00
|
|
|
|
2020-09-13 19:17:58 -07:00
|
|
|
|
2021-03-28 19:08:42 +02:00
|
|
|
|
2021-03-16 00:03:47 -07:00
|
|
|
|
2021-01-08 23:16:50 +01:00
|
|
|
|
2021-07-03 11:47:58 -07:00
|
|
|
|
2022-03-11 19:31:29 -07:00
|
|
|
|
2022-11-25 17:12:50 -07:00
|
|
|
|
2022-12-12 13:16:09 +02:00
|
|
|
|
2023-05-02 15:01:45 -07:00
|
|
|
|
2023-06-20 14:02:04 -04:00
|
|
|
|
2023-06-25 02:26:48 +01:00
|
|
|
|
2023-05-02 15:01:45 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2020-09-13 19:17:58 -07:00
|
|
|
|
|
|
|
|
|
2021-10-29 00:37:25 +01:00
|
|
|
|
2020-09-13 19:17:58 -07:00
|
|
|
|
|
|
|
|
|
stage2: introduce CacheMode
The two CacheMode values are `whole` and `incremental`.
`incremental` is what we had before; `whole` is new.
Whole cache mode uses everything as inputs to the cache hash;
and when a hit occurs it skips everything including linking.
This is ideal for when source files change rarely and for backends that
do not have good incremental compilation support, for example
compiler-rt or libc compiled with LLVM with optimizations on.
This is the main motivation for the additional mode, so that we can have
LLVM-optimized compiler-rt/libc builds, without waiting for the LLVM
backend every single time Zig is invoked.
Incremental cache mode hashes only the input file path and a few target
options, intentionally relying on collisions to locate already-existing
build artifacts which can then be incrementally updated.
The bespoke logic for caching stage1 backend build artifacts
is removed since we now have a global caching mechanism for
when we want to cache the entire compilation, *including* linking.
Previously we had to get "creative" with libs.txt and a special
byte in the hash id to communicate flags, so that when the cached
artifacts were re-linked, we had this information from stage1
even though we didn't actually run it. Now that `CacheMode.whole`
includes linking, this extra information does not need to be
preserved for cache hits. So although this changeset introduces
complexity, it also removes complexity.
The main trickiness here comes from the inherent differences between the
two modes: `incremental` wants a directory immediately to operate on,
while `whole` doesn't know the output directory until the compilation is
complete. This commit deals with this problem mostly inside `update()`,
where, on a cache miss, it replaces `zig_cache_artifact_directory` with a
temporary directory, and then renames it into place once the compilation is
complete.
Items remaining before this branch can be merged:
* [ ] make sure these things make it into the cache manifest:
- @import files
- @embedFile files
- we already add dep files from c but make sure the main .c files make
it in there too, not just the included files
* [ ] double check that the emit paths of other things besides the binary
are working correctly.
* [ ] test `-fno-emit-bin` + `-fstage1`
* [ ] test `-femit-bin=foo` + `-fstage1`
* [ ] implib emit directory copies bin_file_emit directory in create() and needs
to be adjusted to be overridden as well.
* [ ] make sure emit-h is handled correctly in the cache hash
* [ ] Cache: detect duplicate files added to the manifest
Some preliminary performance measurements of wall clock time and
peak RSS used:
stage1 behavior (1077 tests), llvm backend, release build:
* cold global cache: 4.6s, 1.1 GiB
* warm global cache: 3.4s, 980 MiB
stage2 master branch behavior (575 tests), llvm backend, release build:
* cold global cache: 0.62s, 191 MiB
* warm global cache: 0.40s, 128 MiB
stage2 this branch behavior (575 tests), llvm backend, release build:
* cold global cache: 0.62s, 179 MiB
* warm global cache: 0.27s, 90 MiB
2021-12-29 20:49:16 -07:00
|
|
|
|
2020-09-13 19:17:58 -07:00
|
|
|
|
2021-07-23 22:23:03 -07:00
|
|
|
|
2023-10-06 17:05:58 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2022-03-31 19:07:05 -07:00
|
|
|
|
2021-04-25 10:43:07 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-05-14 17:41:22 -07:00
|
|
|
|
2022-04-19 21:51:08 -07:00
|
|
|
|
2021-05-14 17:41:22 -07:00
|
|
|
|
|
|
|
|
|
2022-11-27 01:07:35 -07:00
|
|
|
|
2020-09-13 19:17:58 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2022-11-27 01:07:35 -07:00
|
|
|
|
2021-10-17 18:57:54 -07:00
|
|
|
|
|
|
|
|
|
stage2: entry point via std lib and proper updated file detection
Instead of Module setting up the root_scope with the root source file,
instead, Module relies on the package table graph being set up properly,
and inside `update()`, it does the equivalent of `_ = @import("std");`.
This, in term, imports start.zig, which has the logic to call main (or
not). `Module` no longer has `root_scope` - the root source file is no
longer special, it's just in the package table mapped to "root".
I also went ahead and implemented proper detection of updated files.
mtime, inode, size, and source hash are kept in `Scope.File`.
During an update, iterate over `import_table` and stat each file to find
out which ones are updated.
The source hash is redundant with the source hash used by the struct
decl that corresponds to the file, so it should be removed in a future
commit before merging the branch.
* AstGen: add "previously declared here" notes for variables shadowing
decls.
* Parse imports as structs. Module now calls `AstGen.structDeclInner`,
which is called by `AstGen.containerDecl`.
- `importFile` is a bit kludgy with how it handles the top level Decl
that kinda gets merged into the struct decl at the end of the
function. Be on the look out for bugs related to that as well as
possibly cleaner ways to implement this.
* Module: factor out lookupDeclName into lookupIdentifier and lookupNa
* Rename `Scope.Container` to `Scope.Namespace`.
* Delete some dead code.
This branch won't work until `usingnamespace` is implemented because it
relies on `@import("builtin").OutputMode` and `OutputMode` comes from a
`usingnamespace`.
2021-04-09 23:17:50 -07:00
|
|
|
|
|
|
|
|
|
2021-10-01 17:45:13 -05:00
|
|
|
|
2021-10-17 18:57:54 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
stage2: entry point via std lib and proper updated file detection
Instead of Module setting up the root_scope with the root source file,
instead, Module relies on the package table graph being set up properly,
and inside `update()`, it does the equivalent of `_ = @import("std");`.
This, in term, imports start.zig, which has the logic to call main (or
not). `Module` no longer has `root_scope` - the root source file is no
longer special, it's just in the package table mapped to "root".
I also went ahead and implemented proper detection of updated files.
mtime, inode, size, and source hash are kept in `Scope.File`.
During an update, iterate over `import_table` and stat each file to find
out which ones are updated.
The source hash is redundant with the source hash used by the struct
decl that corresponds to the file, so it should be removed in a future
commit before merging the branch.
* AstGen: add "previously declared here" notes for variables shadowing
decls.
* Parse imports as structs. Module now calls `AstGen.structDeclInner`,
which is called by `AstGen.containerDecl`.
- `importFile` is a bit kludgy with how it handles the top level Decl
that kinda gets merged into the struct decl at the end of the
function. Be on the look out for bugs related to that as well as
possibly cleaner ways to implement this.
* Module: factor out lookupDeclName into lookupIdentifier and lookupNa
* Rename `Scope.Container` to `Scope.Namespace`.
* Delete some dead code.
This branch won't work until `usingnamespace` is implemented because it
relies on `@import("builtin").OutputMode` and `OutputMode` comes from a
`usingnamespace`.
2021-04-09 23:17:50 -07:00
|
|
|
|
2023-07-06 15:27:47 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-05-02 15:01:45 -07:00
|
|
|
|
|
|
|
|
|
2023-09-14 19:59:58 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
compiler: eliminate Decl.value_arena and Sema.perm_arena
The main motivation for this commit is eliminating Decl.value_arena.
Everything else is dominoes.
Decl.name used to be stored in the GPA, now it is stored in InternPool.
It ended up being simpler to migrate other strings to be interned as
well, such as struct field names, union field names, and a few others.
This ended up requiring a big diff, sorry about that. But the changes
are pretty nice, we finally start to take advantage of InternPool's
existence.
global_error_set and error_name_list are simplified. Now it is a single
ArrayHashMap(NullTerminatedString, void) and the index is the error tag
value.
Module.tmp_hack_arena is re-introduced (it was removed in
eeff407941560ce8eb5b737b2436dfa93cfd3a0c) in order to deal with
comptime_args, optimized_order, and struct and union fields. After
structs and unions get moved into InternPool properly, tmp_hack_arena
can be deleted again.
2023-06-01 23:46:04 -07:00
|
|
|
|
|
|
|
|
|
2023-08-21 14:27:34 -07:00
|
|
|
|
compiler: eliminate Decl.value_arena and Sema.perm_arena
The main motivation for this commit is eliminating Decl.value_arena.
Everything else is dominoes.
Decl.name used to be stored in the GPA, now it is stored in InternPool.
It ended up being simpler to migrate other strings to be interned as
well, such as struct field names, union field names, and a few others.
This ended up requiring a big diff, sorry about that. But the changes
are pretty nice, we finally start to take advantage of InternPool's
existence.
global_error_set and error_name_list are simplified. Now it is a single
ArrayHashMap(NullTerminatedString, void) and the index is the error tag
value.
Module.tmp_hack_arena is re-introduced (it was removed in
eeff407941560ce8eb5b737b2436dfa93cfd3a0c) in order to deal with
comptime_args, optimized_order, and struct and union fields. After
structs and unions get moved into InternPool properly, tmp_hack_arena
can be deleted again.
2023-06-01 23:46:04 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2020-09-13 19:17:58 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2022-04-19 21:51:08 -07:00
|
|
|
|
2021-01-16 22:51:01 -07:00
|
|
|
|
2021-05-14 23:10:38 -07:00
|
|
|
|
2022-04-19 21:51:08 -07:00
|
|
|
|
2020-09-13 19:17:58 -07:00
|
|
|
|
2021-10-01 17:45:13 -05:00
|
|
|
|
|
|
|
|
|
2021-10-17 18:57:54 -07:00
|
|
|
|
|
|
|
|
|
2020-09-13 19:17:58 -07:00
|
|
|
|
|
|
|
|
|
2021-01-16 22:51:01 -07:00
|
|
|
|
2022-12-12 13:16:09 +02:00
|
|
|
|
|
|
|
|
|
2023-09-25 04:54:07 +02:00
|
|
|
|
2020-09-13 19:17:58 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2022-04-19 21:51:08 -07:00
|
|
|
|
2020-09-13 19:17:58 -07:00
|
|
|
|
compiler: eliminate Decl.value_arena and Sema.perm_arena
The main motivation for this commit is eliminating Decl.value_arena.
Everything else is dominoes.
Decl.name used to be stored in the GPA, now it is stored in InternPool.
It ended up being simpler to migrate other strings to be interned as
well, such as struct field names, union field names, and a few others.
This ended up requiring a big diff, sorry about that. But the changes
are pretty nice, we finally start to take advantage of InternPool's
existence.
global_error_set and error_name_list are simplified. Now it is a single
ArrayHashMap(NullTerminatedString, void) and the index is the error tag
value.
Module.tmp_hack_arena is re-introduced (it was removed in
eeff407941560ce8eb5b737b2436dfa93cfd3a0c) in order to deal with
comptime_args, optimized_order, and struct and union fields. After
structs and unions get moved into InternPool properly, tmp_hack_arena
can be deleted again.
2023-06-01 23:46:04 -07:00
|
|
|
|
|
|
|
|
|
2021-03-26 17:54:41 -04:00
|
|
|
|
2023-01-24 13:35:10 +02:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2020-09-13 19:17:58 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2020-09-28 15:42:09 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2020-09-28 00:06:06 -07:00
|
|
|
|
2021-04-22 22:35:18 -07:00
|
|
|
|
|
|
|
|
|
2021-03-25 23:00:38 -07:00
|
|
|
|
2021-01-16 22:51:01 -07:00
|
|
|
|
2021-04-26 20:41:07 -07:00
|
|
|
|
|
|
|
|
|
2022-04-19 21:51:08 -07:00
|
|
|
|
|
|
|
|
|
2022-05-04 20:38:53 -07:00
|
|
|
|
|
|
|
|
|
2022-09-09 18:55:58 +03:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-06-24 15:03:15 -07:00
|
|
|
|
rework generic function calls
Abridged summary:
* Move `Module.Fn` into `InternPool`.
* Delete a lot of confusing and problematic `Sema` logic related to
generic function calls.
This commit removes `Module.Fn` and replaces it with two new
`InternPool.Tag` values:
* `func_decl` - corresponding to a function declared in the source
code. This one contains line/column numbers, zir_body_inst, etc.
* `func_instance` - one for each monomorphization of a generic
function. Contains a reference to the `func_decl` from whence the
instantiation came, along with the `comptime` parameter values (or
types in the case of `anytype`)
Since `InternPool` provides deduplication on these values, these fields
are now deleted from `Module`:
* `monomorphed_func_keys`
* `monomorphed_funcs`
* `align_stack_fns`
Instead of these, Sema logic for generic function instantiation now
unconditionally evaluates the function prototype expression for every
generic callsite. This is technically required in order for type
coercions to work. The previous code had some dubious, probably wrong
hacks to make things work, such as `hashUncoerced`. I'm not 100% sure
how we were able to eliminate that function and still pass all the
behavior tests, but I'm pretty sure things were still broken without
doing type coercion for every generic function call argument.
After the function prototype is evaluated, it produces a deduplicated
`func_instance` `InternPool.Index` which can then be used for the
generic function call.
Some other nice things made by this simplification are the removal of
`comptime_args_fn_inst` and `preallocated_new_func` from `Sema`, and the
messy logic associated with them.
I have not yet been able to measure the perf of this against master
branch. On one hand, it reduces memory usage and pointer chasing of the
most heavily used `InternPool` Tag - function bodies - but on the other
hand, it does evaluate function prototype expressions more than before.
We will soon find out.
2023-07-03 22:09:30 -07:00
|
|
|
|
|
|
|
|
|
2023-06-24 15:03:15 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
compiler: eliminate Decl.value_arena and Sema.perm_arena
The main motivation for this commit is eliminating Decl.value_arena.
Everything else is dominoes.
Decl.name used to be stored in the GPA, now it is stored in InternPool.
It ended up being simpler to migrate other strings to be interned as
well, such as struct field names, union field names, and a few others.
This ended up requiring a big diff, sorry about that. But the changes
are pretty nice, we finally start to take advantage of InternPool's
existence.
global_error_set and error_name_list are simplified. Now it is a single
ArrayHashMap(NullTerminatedString, void) and the index is the error tag
value.
Module.tmp_hack_arena is re-introduced (it was removed in
eeff407941560ce8eb5b737b2436dfa93cfd3a0c) in order to deal with
comptime_args, optimized_order, and struct and union fields. After
structs and unions get moved into InternPool properly, tmp_hack_arena
can be deleted again.
2023-06-01 23:46:04 -07:00
|
|
|
|
|
|
|
|
|
2022-12-12 13:16:09 +02:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-04-26 20:41:07 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2022-04-19 21:51:08 -07:00
|
|
|
|
2021-04-26 20:41:07 -07:00
|
|
|
|
2022-04-19 21:51:08 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-06-15 13:14:16 +06:00
|
|
|
|
2022-04-19 21:51:08 -07:00
|
|
|
|
2021-04-26 20:41:07 -07:00
|
|
|
|
|
|
|
|
|
2021-03-28 19:38:19 -07:00
|
|
|
|
|
|
|
|
|
2020-09-13 19:17:58 -07:00
|
|
|
|
2023-06-12 01:44:12 -04:00
|
|
|
|
stage2: *WIP*: rework ZIR memory layout; overhaul source locations
The memory layout for ZIR instructions is completely reworked. See
zir.zig for those changes. Some new types:
* `zir.Code`: a "finished" set of ZIR instructions. Instead of allocating
each instruction independently, there is now a Tag and 8 bytes of
data available for all ZIR instructions. Small instructions fit
within these 8 bytes; larger ones use 4 bytes for an index into
`extra`. There is also `string_bytes` so that we can have 4 byte
references to strings. `zir.Inst.Tag` describes how to interpret
those 8 bytes of data.
- This is shared by all `Block` scopes.
* `Module.WipZirCode`: represents an in-progress `zir.Code`. In this
structure, the arrays are mutable, and get resized as we add/delete
things. There is extra state to keep track of things. This struct is
stored on the stack. Once it is finished, it produces an immutable
`zir.Code`, which will remain on the heap for the duration of a
function's existence.
- This is shared by all `GenZir` scopes.
* `Sema`: represents in-progress semantic analysis of a `zir.Code`.
This data is stored on the stack and is shared among all `Block`
scopes. It is now the main "self" argument to everything in the file
that was previously named `zir_sema.zig`.
Additionally, I moved some logic that was in `Module` into here.
`Module.Fn` now stores its parameter names inside the `zir.Code`,
instead of inside ZIR instructions. When the TZIR memory layout
reworking time comes, codegen will be able to reference this data
directly instead of duplicating it.
astgen.zig is (so far) almost entirely untouched, but nearly all of it
will need to be reworked to adhere to this new memory layout structure.
I have no benchmarks to report yet, as I am still working through
compile errors and fixing various things that I broke in this branch.
Overhaul of Source Locations:
Previously we used `usize` everywhere to mean byte offset, but sometimes
also mean other stuff. This was error prone and also made us do
unnecessary work, and store unnecessary bytes in memory.
Now there are more types involved into source locations, and more ways
to describe a source location.
* AllErrors.Message: embrace the assumption that files always have less
than 2 << 32 bytes.
* SrcLoc gets more complicated, to model more complicated source
locations.
* Introduce LazySrcLoc, which can model interesting source locations
with very little stored state. Useful for avoiding doing unnecessary
work when no compile errors occur.
Also, previously, we had `src: usize` on every ZIR instruction. This is
no longer the case. Each instruction now determines whether it even cares
about source location, and if so, how that source location is stored.
This requires more careful work inside `Sema`, but it results in fewer
bytes stored on the heap, without compromising accuracy and power of
compile error messages.
Miscellaneous:
* std.zig: string literals have more helpful result values for
reporting errors. There is now a lower level API and a higher level
API.
- side note: I noticed that the string literal logic needs some love.
There is some unnecessarily hacky code there.
* cut & pasted some TZIR logic that was in zir.zig to ir.zig. This
probably broke stuff and needs to get fixed.
* Removed type/Enum.zig, type/Union.zig, and type/Struct.zig. I don't
think this quite how this code will be organized. Need some more
careful planning about how to implement structs, unions, enums. They
need to be independent Decls, just like a top level function.
2021-03-15 23:38:38 -07:00
|
|
|
|
2020-09-13 19:17:58 -07:00
|
|
|
|
2022-04-19 21:51:08 -07:00
|
|
|
|
2021-09-23 18:09:35 -05:00
|
|
|
|
|
|
|
|
|
2022-04-19 21:51:08 -07:00
|
|
|
|
2020-09-13 19:17:58 -07:00
|
|
|
|
2022-04-19 21:51:08 -07:00
|
|
|
|
2020-09-13 19:17:58 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-05-14 17:41:22 -07:00
|
|
|
|
2023-06-12 01:44:12 -04:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2022-04-19 21:51:08 -07:00
|
|
|
|
|
|
|
|
|
2021-05-14 17:41:22 -07:00
|
|
|
|
2023-05-08 16:52:59 -07:00
|
|
|
|
2022-04-19 21:51:08 -07:00
|
|
|
|
2021-05-14 17:41:22 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2020-09-13 19:17:58 -07:00
|
|
|
|
|
|
|
|
|
2021-09-23 12:17:06 -05:00
|
|
|
|
2023-09-14 19:59:58 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-06-01 08:10:45 +01:00
|
|
|
|
|
|
|
|
|
2023-09-14 19:59:58 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2022-09-07 22:05:01 +03:00
|
|
|
|
2023-09-14 19:59:58 -07:00
|
|
|
|
|
|
|
|
|
2023-06-01 08:10:45 +01:00
|
|
|
|
2023-09-14 19:59:58 -07:00
|
|
|
|
2021-09-23 12:17:06 -05:00
|
|
|
|
|
|
|
|
|
2023-09-14 19:59:58 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-09-23 12:17:06 -05:00
|
|
|
|
2023-04-23 19:09:12 -04:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-04-27 01:11:41 -04:00
|
|
|
|
|
|
|
|
|
2023-04-23 19:09:12 -04:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-04-27 01:11:41 -04:00
|
|
|
|
2023-04-23 19:09:12 -04:00
|
|
|
|
|
|
|
|
|
2023-04-27 01:11:41 -04:00
|
|
|
|
2023-04-23 19:09:12 -04:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-04-27 01:11:41 -04:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-04-23 19:09:12 -04:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-04-27 01:11:41 -04:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-04-23 19:09:12 -04:00
|
|
|
|
|
|
|
|
|
2020-09-13 19:17:58 -07:00
|
|
|
|
compiler: eliminate Decl.value_arena and Sema.perm_arena
The main motivation for this commit is eliminating Decl.value_arena.
Everything else is dominoes.
Decl.name used to be stored in the GPA, now it is stored in InternPool.
It ended up being simpler to migrate other strings to be interned as
well, such as struct field names, union field names, and a few others.
This ended up requiring a big diff, sorry about that. But the changes
are pretty nice, we finally start to take advantage of InternPool's
existence.
global_error_set and error_name_list are simplified. Now it is a single
ArrayHashMap(NullTerminatedString, void) and the index is the error tag
value.
Module.tmp_hack_arena is re-introduced (it was removed in
eeff407941560ce8eb5b737b2436dfa93cfd3a0c) in order to deal with
comptime_args, optimized_order, and struct and union fields. After
structs and unions get moved into InternPool properly, tmp_hack_arena
can be deleted again.
2023-06-01 23:46:04 -07:00
|
|
|
|
2021-04-27 18:36:12 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
compiler: eliminate Decl.value_arena and Sema.perm_arena
The main motivation for this commit is eliminating Decl.value_arena.
Everything else is dominoes.
Decl.name used to be stored in the GPA, now it is stored in InternPool.
It ended up being simpler to migrate other strings to be interned as
well, such as struct field names, union field names, and a few others.
This ended up requiring a big diff, sorry about that. But the changes
are pretty nice, we finally start to take advantage of InternPool's
existence.
global_error_set and error_name_list are simplified. Now it is a single
ArrayHashMap(NullTerminatedString, void) and the index is the error tag
value.
Module.tmp_hack_arena is re-introduced (it was removed in
eeff407941560ce8eb5b737b2436dfa93cfd3a0c) in order to deal with
comptime_args, optimized_order, and struct and union fields. After
structs and unions get moved into InternPool properly, tmp_hack_arena
can be deleted again.
2023-06-01 23:46:04 -07:00
|
|
|
|
2021-04-27 18:36:12 -07:00
|
|
|
|
2023-06-20 14:02:04 -04:00
|
|
|
|
2021-09-02 14:46:31 +02:00
|
|
|
|
|
|
|
|
|
stage2: entry point via std lib and proper updated file detection
Instead of Module setting up the root_scope with the root source file,
instead, Module relies on the package table graph being set up properly,
and inside `update()`, it does the equivalent of `_ = @import("std");`.
This, in term, imports start.zig, which has the logic to call main (or
not). `Module` no longer has `root_scope` - the root source file is no
longer special, it's just in the package table mapped to "root".
I also went ahead and implemented proper detection of updated files.
mtime, inode, size, and source hash are kept in `Scope.File`.
During an update, iterate over `import_table` and stat each file to find
out which ones are updated.
The source hash is redundant with the source hash used by the struct
decl that corresponds to the file, so it should be removed in a future
commit before merging the branch.
* AstGen: add "previously declared here" notes for variables shadowing
decls.
* Parse imports as structs. Module now calls `AstGen.structDeclInner`,
which is called by `AstGen.containerDecl`.
- `importFile` is a bit kludgy with how it handles the top level Decl
that kinda gets merged into the struct decl at the end of the
function. Be on the look out for bugs related to that as well as
possibly cleaner ways to implement this.
* Module: factor out lookupDeclName into lookupIdentifier and lookupNa
* Rename `Scope.Container` to `Scope.Namespace`.
* Delete some dead code.
This branch won't work until `usingnamespace` is implemented because it
relies on `@import("builtin").OutputMode` and `OutputMode` comes from a
`usingnamespace`.
2021-04-09 23:17:50 -07:00
|
|
|
|
2020-09-13 19:17:58 -07:00
|
|
|
|
2021-05-04 11:08:40 -07:00
|
|
|
|
|
|
|
|
|
2023-05-08 16:52:59 -07:00
|
|
|
|
2021-04-08 20:37:19 -07:00
|
|
|
|
2021-09-23 12:17:06 -05:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-09-14 19:59:58 -07:00
|
|
|
|
2021-09-23 12:17:06 -05:00
|
|
|
|
2021-04-08 20:37:19 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
stage2: entry point via std lib and proper updated file detection
Instead of Module setting up the root_scope with the root source file,
instead, Module relies on the package table graph being set up properly,
and inside `update()`, it does the equivalent of `_ = @import("std");`.
This, in term, imports start.zig, which has the logic to call main (or
not). `Module` no longer has `root_scope` - the root source file is no
longer special, it's just in the package table mapped to "root".
I also went ahead and implemented proper detection of updated files.
mtime, inode, size, and source hash are kept in `Scope.File`.
During an update, iterate over `import_table` and stat each file to find
out which ones are updated.
The source hash is redundant with the source hash used by the struct
decl that corresponds to the file, so it should be removed in a future
commit before merging the branch.
* AstGen: add "previously declared here" notes for variables shadowing
decls.
* Parse imports as structs. Module now calls `AstGen.structDeclInner`,
which is called by `AstGen.containerDecl`.
- `importFile` is a bit kludgy with how it handles the top level Decl
that kinda gets merged into the struct decl at the end of the
function. Be on the look out for bugs related to that as well as
possibly cleaner ways to implement this.
* Module: factor out lookupDeclName into lookupIdentifier and lookupNa
* Rename `Scope.Container` to `Scope.Namespace`.
* Delete some dead code.
This branch won't work until `usingnamespace` is implemented because it
relies on `@import("builtin").OutputMode` and `OutputMode` comes from a
`usingnamespace`.
2021-04-09 23:17:50 -07:00
|
|
|
|
2020-09-13 19:17:58 -07:00
|
|
|
|
2021-08-30 19:22:04 -07:00
|
|
|
|
2021-05-01 21:57:52 -07:00
|
|
|
|
|
|
|
|
|
2022-01-31 17:07:45 +01:00
|
|
|
|
2021-05-01 21:57:52 -07:00
|
|
|
|
2021-04-28 16:55:22 -07:00
|
|
|
|
|
|
|
|
|
2021-05-01 21:57:52 -07:00
|
|
|
|
2021-05-05 13:16:14 -07:00
|
|
|
|
2021-04-28 16:55:22 -07:00
|
|
|
|
2021-04-08 20:37:19 -07:00
|
|
|
|
2020-09-13 19:17:58 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-05-11 17:34:13 -07:00
|
|
|
|
|
|
|
|
|
2020-09-13 19:17:58 -07:00
|
|
|
|
2021-05-11 17:34:13 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2020-09-13 19:17:58 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-04-10 00:00:35 -04:00
|
|
|
|
|
|
|
|
|
2020-09-13 19:17:58 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2022-03-20 00:35:09 -07:00
|
|
|
|
2021-04-27 18:36:12 -07:00
|
|
|
|
2021-05-11 14:17:52 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-04-07 19:38:00 -07:00
|
|
|
|
2020-09-13 19:17:58 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-04-27 18:36:12 -07:00
|
|
|
|
|
|
|
|
|
2021-04-28 16:55:22 -07:00
|
|
|
|
|
|
|
|
|
2021-09-02 14:46:31 +02:00
|
|
|
|
|
|
|
|
|
stage2: garbage collect unused anon decls
After this change, the frontend and backend cooperate to keep track of
which Decls are actually emitted into the machine code. When any backend
sees a `decl_ref` Value, it must mark the corresponding Decl `alive`
field to true.
This prevents unused comptime data from spilling into the output object
files. For example, if you do an `inline for` loop, previously, any
intermediate value calculations would have gone into the object file.
Now they are garbage collected immediately after the owner Decl has its
machine code generated.
In the frontend, when it is time to send a Decl to the linker, if it has
not been marked "alive" then it is deleted instead.
Additional improvements:
* Resolve type ABI layouts after successful semantic analysis of a
Decl. This is needed so that the backend has access to struct fields.
* Sema: fix incorrect logic in resolveMaybeUndefVal. It should return
"not comptime known" instead of a compile error for global variables.
* `Value.pointerDeref` now returns `null` in the case that the pointer
deref cannot happen at compile-time. This is true for global
variables, for example. Another example is if a comptime known
pointer has a hard coded address value.
* Binary arithmetic sets the requireRuntimeBlock source location to the
lhs_src or rhs_src as appropriate instead of on the operator node.
* Fix LLVM codegen for slice_elem_val which had the wrong logic for
when the operand was not a pointer.
As noted in the comment in the implementation of deleteUnusedDecl, a
future improvement will be to rework the frontend/linker interface to
remove the frontend's responsibility of calling allocateDeclIndexes.
I discovered some issues with the plan9 linker backend that are related
to this, and worked around them for now.
2021-07-29 19:30:37 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2022-06-28 17:15:10 +03:00
|
|
|
|
|
|
|
|
|
2023-01-10 21:54:58 +02:00
|
|
|
|
|
|
|
|
|
2020-09-13 19:17:58 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-01-10 21:54:58 +02:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2022-04-19 21:51:08 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-06-22 18:46:56 +01:00
|
|
|
|
2022-04-19 21:51:08 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-06-22 18:46:56 +01:00
|
|
|
|
2022-04-19 21:51:08 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-06-22 18:46:56 +01:00
|
|
|
|
2022-04-19 21:51:08 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-04-19 17:40:29 +01:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2020-09-13 19:17:58 -07:00
|
|
|
|
2021-04-28 16:55:22 -07:00
|
|
|
|
|
|
|
|
|
2023-05-08 16:52:59 -07:00
|
|
|
|
|
|
|
|
|
2021-04-28 16:55:22 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-05-05 13:16:14 -07:00
|
|
|
|
2021-05-01 21:57:52 -07:00
|
|
|
|
2021-04-28 16:55:22 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-05-08 16:52:59 -07:00
|
|
|
|
|
|
|
|
|
2021-04-28 16:55:22 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-05-05 13:16:14 -07:00
|
|
|
|
2021-04-28 16:55:22 -07:00
|
|
|
|
2023-06-22 18:46:56 +01:00
|
|
|
|
2021-04-28 16:55:22 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-05-08 16:52:59 -07:00
|
|
|
|
2021-05-05 13:16:14 -07:00
|
|
|
|
2023-05-08 16:52:59 -07:00
|
|
|
|
2021-05-01 21:57:52 -07:00
|
|
|
|
2021-04-28 16:55:22 -07:00
|
|
|
|
|
|
|
|
|
2023-05-08 16:52:59 -07:00
|
|
|
|
2021-04-28 16:55:22 -07:00
|
|
|
|
2021-05-05 13:16:14 -07:00
|
|
|
|
2023-05-08 16:52:59 -07:00
|
|
|
|
2023-06-22 18:46:56 +01:00
|
|
|
|
2021-04-28 16:55:22 -07:00
|
|
|
|
|
|
|
|
|
2023-05-08 16:52:59 -07:00
|
|
|
|
2021-09-02 14:46:31 +02:00
|
|
|
|
2021-05-05 13:16:14 -07:00
|
|
|
|
2023-05-08 16:52:59 -07:00
|
|
|
|
2023-06-15 13:14:16 +06:00
|
|
|
|
2023-06-22 18:46:56 +01:00
|
|
|
|
2021-09-02 14:46:31 +02:00
|
|
|
|
|
|
|
|
|
2023-05-08 16:52:59 -07:00
|
|
|
|
2021-09-02 14:46:31 +02:00
|
|
|
|
|
|
|
|
|
2023-05-08 16:52:59 -07:00
|
|
|
|
2023-06-15 13:14:16 +06:00
|
|
|
|
2023-06-22 18:46:56 +01:00
|
|
|
|
2021-04-28 16:55:22 -07:00
|
|
|
|
|
|
|
|
|
2021-05-01 21:57:52 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-08-30 19:22:04 -07:00
|
|
|
|
2023-06-22 18:46:56 +01:00
|
|
|
|
2021-03-20 17:09:06 -07:00
|
|
|
|
|
|
|
|
|
2021-08-30 19:22:04 -07:00
|
|
|
|
2023-06-22 18:46:56 +01:00
|
|
|
|
2021-03-20 17:09:06 -07:00
|
|
|
|
|
|
|
|
|
2021-08-30 19:22:04 -07:00
|
|
|
|
2021-03-19 14:59:46 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-08-30 19:22:04 -07:00
|
|
|
|
2022-06-07 17:48:53 -07:00
|
|
|
|
2021-03-19 23:06:19 -07:00
|
|
|
|
|
|
|
|
|
2023-05-08 16:52:59 -07:00
|
|
|
|
|
|
|
|
|
2021-05-14 23:10:38 -07:00
|
|
|
|
|
|
|
|
|
2023-05-08 16:52:59 -07:00
|
|
|
|
2021-01-16 22:51:01 -07:00
|
|
|
|
2023-05-08 16:52:59 -07:00
|
|
|
|
2021-04-16 14:44:02 -07:00
|
|
|
|
2022-06-07 17:48:53 -07:00
|
|
|
|
2021-01-16 22:51:01 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-05-08 16:52:59 -07:00
|
|
|
|
|
|
|
|
|
2021-04-08 20:37:19 -07:00
|
|
|
|
2020-09-13 19:17:58 -07:00
|
|
|
|
|
|
|
|
|
2023-05-08 16:52:59 -07:00
|
|
|
|
|
|
|
|
|
stage2: *WIP*: rework ZIR memory layout; overhaul source locations
The memory layout for ZIR instructions is completely reworked. See
zir.zig for those changes. Some new types:
* `zir.Code`: a "finished" set of ZIR instructions. Instead of allocating
each instruction independently, there is now a Tag and 8 bytes of
data available for all ZIR instructions. Small instructions fit
within these 8 bytes; larger ones use 4 bytes for an index into
`extra`. There is also `string_bytes` so that we can have 4 byte
references to strings. `zir.Inst.Tag` describes how to interpret
those 8 bytes of data.
- This is shared by all `Block` scopes.
* `Module.WipZirCode`: represents an in-progress `zir.Code`. In this
structure, the arrays are mutable, and get resized as we add/delete
things. There is extra state to keep track of things. This struct is
stored on the stack. Once it is finished, it produces an immutable
`zir.Code`, which will remain on the heap for the duration of a
function's existence.
- This is shared by all `GenZir` scopes.
* `Sema`: represents in-progress semantic analysis of a `zir.Code`.
This data is stored on the stack and is shared among all `Block`
scopes. It is now the main "self" argument to everything in the file
that was previously named `zir_sema.zig`.
Additionally, I moved some logic that was in `Module` into here.
`Module.Fn` now stores its parameter names inside the `zir.Code`,
instead of inside ZIR instructions. When the TZIR memory layout
reworking time comes, codegen will be able to reference this data
directly instead of duplicating it.
astgen.zig is (so far) almost entirely untouched, but nearly all of it
will need to be reworked to adhere to this new memory layout structure.
I have no benchmarks to report yet, as I am still working through
compile errors and fixing various things that I broke in this branch.
Overhaul of Source Locations:
Previously we used `usize` everywhere to mean byte offset, but sometimes
also mean other stuff. This was error prone and also made us do
unnecessary work, and store unnecessary bytes in memory.
Now there are more types involved into source locations, and more ways
to describe a source location.
* AllErrors.Message: embrace the assumption that files always have less
than 2 << 32 bytes.
* SrcLoc gets more complicated, to model more complicated source
locations.
* Introduce LazySrcLoc, which can model interesting source locations
with very little stored state. Useful for avoiding doing unnecessary
work when no compile errors occur.
Also, previously, we had `src: usize` on every ZIR instruction. This is
no longer the case. Each instruction now determines whether it even cares
about source location, and if so, how that source location is stored.
This requires more careful work inside `Sema`, but it results in fewer
bytes stored on the heap, without compromising accuracy and power of
compile error messages.
Miscellaneous:
* std.zig: string literals have more helpful result values for
reporting errors. There is now a lower level API and a higher level
API.
- side note: I noticed that the string literal logic needs some love.
There is some unnecessarily hacky code there.
* cut & pasted some TZIR logic that was in zir.zig to ir.zig. This
probably broke stuff and needs to get fixed.
* Removed type/Enum.zig, type/Union.zig, and type/Struct.zig. I don't
think this quite how this code will be organized. Need some more
careful planning about how to implement structs, unions, enums. They
need to be independent Decls, just like a top level function.
2021-03-15 23:38:38 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2022-04-19 21:51:08 -07:00
|
|
|
|
2022-06-28 17:15:10 +03:00
|
|
|
|
2023-06-12 01:44:12 -04:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2022-06-28 17:15:10 +03:00
|
|
|
|
stage2: implement simple enums
A simple enum is an enum which has an automatic integer tag type,
all tag values automatically assigned, and no top level declarations.
Such enums are created directly in AstGen and shared by all the
generic/comptime instantiations of the surrounding ZIR code. This
commit implements, but does not yet add any test cases for, simple enums.
A full enum is an enum for which any of the above conditions are not
true. Full enums are created in Sema, and therefore will create a unique
type per generic/comptime instantiation. This commit does not implement
full enums. However the `enum_decl_nonexhaustive` ZIR instruction is
added and the respective Type functions are filled out.
This commit makes an improvement to ZIR code, removing the decls array
and removing the decl_map from AstGen. Instead, decl_ref and
decl_val ZIR instructions index into the `owner_decl.dependencies`
ArrayHashMap. We already need this dependencies array for incremental
compilation purposes, and so repurposing it to also use it for ZIR decl
indexes makes for efficient memory usage.
Similarly, this commit fixes up incorrect memory management by removing
the `const` ZIR instruction. The two places it was used stored memory in
the AstGen arena, which may get freed after Sema. Now it properly sets
up a new anonymous Decl for error sets and uses a normal decl_val
instruction.
The other usage of `const` ZIR instruction was float literals. These are
now changed to use `float` ZIR instruction when the value fits inside
`zir.Inst.Data` and `float128` otherwise.
AstGen + Sema: implement int_to_enum and enum_to_int. No tests yet; I expect to
have to make some fixes before they will pass tests. Will do that in the
branch before merging.
AstGen: fix struct astgen incorrectly counting decls as fields.
Type/Value: give up on trying to exhaustively list every tag all the
time. This makes the file more manageable. Also found a bug with
i128/u128 this way, since the name of the function was more obvious when
looking at the tag values.
Type: implement abiAlignment and abiSize for structs. This will need to
get more sophisticated at some point, but for now it is progress.
Value: add new `enum_field_index` tag.
Value: add hash_u32, needed when using ArrayHashMap.
2021-04-06 17:43:56 -07:00
|
|
|
|
|
|
|
|
|
2022-04-19 21:51:08 -07:00
|
|
|
|
2023-06-12 01:44:12 -04:00
|
|
|
|
2021-09-30 21:43:30 -05:00
|
|
|
|
|
|
|
|
|
compiler: eliminate Decl.value_arena and Sema.perm_arena
The main motivation for this commit is eliminating Decl.value_arena.
Everything else is dominoes.
Decl.name used to be stored in the GPA, now it is stored in InternPool.
It ended up being simpler to migrate other strings to be interned as
well, such as struct field names, union field names, and a few others.
This ended up requiring a big diff, sorry about that. But the changes
are pretty nice, we finally start to take advantage of InternPool's
existence.
global_error_set and error_name_list are simplified. Now it is a single
ArrayHashMap(NullTerminatedString, void) and the index is the error tag
value.
Module.tmp_hack_arena is re-introduced (it was removed in
eeff407941560ce8eb5b737b2436dfa93cfd3a0c) in order to deal with
comptime_args, optimized_order, and struct and union fields. After
structs and unions get moved into InternPool properly, tmp_hack_arena
can be deleted again.
2023-06-01 23:46:04 -07:00
|
|
|
|
2023-06-02 00:33:28 -07:00
|
|
|
|
|
|
|
|
|
compiler: eliminate Decl.value_arena and Sema.perm_arena
The main motivation for this commit is eliminating Decl.value_arena.
Everything else is dominoes.
Decl.name used to be stored in the GPA, now it is stored in InternPool.
It ended up being simpler to migrate other strings to be interned as
well, such as struct field names, union field names, and a few others.
This ended up requiring a big diff, sorry about that. But the changes
are pretty nice, we finally start to take advantage of InternPool's
existence.
global_error_set and error_name_list are simplified. Now it is a single
ArrayHashMap(NullTerminatedString, void) and the index is the error tag
value.
Module.tmp_hack_arena is re-introduced (it was removed in
eeff407941560ce8eb5b737b2436dfa93cfd3a0c) in order to deal with
comptime_args, optimized_order, and struct and union fields. After
structs and unions get moved into InternPool properly, tmp_hack_arena
can be deleted again.
2023-06-01 23:46:04 -07:00
|
|
|
|
2023-06-02 00:33:28 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-06-12 01:44:12 -04:00
|
|
|
|
2023-06-02 00:33:28 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
compiler: eliminate Decl.value_arena and Sema.perm_arena
The main motivation for this commit is eliminating Decl.value_arena.
Everything else is dominoes.
Decl.name used to be stored in the GPA, now it is stored in InternPool.
It ended up being simpler to migrate other strings to be interned as
well, such as struct field names, union field names, and a few others.
This ended up requiring a big diff, sorry about that. But the changes
are pretty nice, we finally start to take advantage of InternPool's
existence.
global_error_set and error_name_list are simplified. Now it is a single
ArrayHashMap(NullTerminatedString, void) and the index is the error tag
value.
Module.tmp_hack_arena is re-introduced (it was removed in
eeff407941560ce8eb5b737b2436dfa93cfd3a0c) in order to deal with
comptime_args, optimized_order, and struct and union fields. After
structs and unions get moved into InternPool properly, tmp_hack_arena
can be deleted again.
2023-06-01 23:46:04 -07:00
|
|
|
|
2023-06-02 00:33:28 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2022-09-16 22:21:14 +02:00
|
|
|
|
|
|
|
|
|
compiler: eliminate Decl.value_arena and Sema.perm_arena
The main motivation for this commit is eliminating Decl.value_arena.
Everything else is dominoes.
Decl.name used to be stored in the GPA, now it is stored in InternPool.
It ended up being simpler to migrate other strings to be interned as
well, such as struct field names, union field names, and a few others.
This ended up requiring a big diff, sorry about that. But the changes
are pretty nice, we finally start to take advantage of InternPool's
existence.
global_error_set and error_name_list are simplified. Now it is a single
ArrayHashMap(NullTerminatedString, void) and the index is the error tag
value.
Module.tmp_hack_arena is re-introduced (it was removed in
eeff407941560ce8eb5b737b2436dfa93cfd3a0c) in order to deal with
comptime_args, optimized_order, and struct and union fields. After
structs and unions get moved into InternPool properly, tmp_hack_arena
can be deleted again.
2023-06-01 23:46:04 -07:00
|
|
|
|
|
|
|
|
|
2022-09-16 22:21:14 +02:00
|
|
|
|
compiler: eliminate Decl.value_arena and Sema.perm_arena
The main motivation for this commit is eliminating Decl.value_arena.
Everything else is dominoes.
Decl.name used to be stored in the GPA, now it is stored in InternPool.
It ended up being simpler to migrate other strings to be interned as
well, such as struct field names, union field names, and a few others.
This ended up requiring a big diff, sorry about that. But the changes
are pretty nice, we finally start to take advantage of InternPool's
existence.
global_error_set and error_name_list are simplified. Now it is a single
ArrayHashMap(NullTerminatedString, void) and the index is the error tag
value.
Module.tmp_hack_arena is re-introduced (it was removed in
eeff407941560ce8eb5b737b2436dfa93cfd3a0c) in order to deal with
comptime_args, optimized_order, and struct and union fields. After
structs and unions get moved into InternPool properly, tmp_hack_arena
can be deleted again.
2023-06-01 23:46:04 -07:00
|
|
|
|
2022-09-16 22:21:14 +02:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
compiler: eliminate Decl.value_arena and Sema.perm_arena
The main motivation for this commit is eliminating Decl.value_arena.
Everything else is dominoes.
Decl.name used to be stored in the GPA, now it is stored in InternPool.
It ended up being simpler to migrate other strings to be interned as
well, such as struct field names, union field names, and a few others.
This ended up requiring a big diff, sorry about that. But the changes
are pretty nice, we finally start to take advantage of InternPool's
existence.
global_error_set and error_name_list are simplified. Now it is a single
ArrayHashMap(NullTerminatedString, void) and the index is the error tag
value.
Module.tmp_hack_arena is re-introduced (it was removed in
eeff407941560ce8eb5b737b2436dfa93cfd3a0c) in order to deal with
comptime_args, optimized_order, and struct and union fields. After
structs and unions get moved into InternPool properly, tmp_hack_arena
can be deleted again.
2023-06-01 23:46:04 -07:00
|
|
|
|
stage2: implement simple enums
A simple enum is an enum which has an automatic integer tag type,
all tag values automatically assigned, and no top level declarations.
Such enums are created directly in AstGen and shared by all the
generic/comptime instantiations of the surrounding ZIR code. This
commit implements, but does not yet add any test cases for, simple enums.
A full enum is an enum for which any of the above conditions are not
true. Full enums are created in Sema, and therefore will create a unique
type per generic/comptime instantiation. This commit does not implement
full enums. However the `enum_decl_nonexhaustive` ZIR instruction is
added and the respective Type functions are filled out.
This commit makes an improvement to ZIR code, removing the decls array
and removing the decl_map from AstGen. Instead, decl_ref and
decl_val ZIR instructions index into the `owner_decl.dependencies`
ArrayHashMap. We already need this dependencies array for incremental
compilation purposes, and so repurposing it to also use it for ZIR decl
indexes makes for efficient memory usage.
Similarly, this commit fixes up incorrect memory management by removing
the `const` ZIR instruction. The two places it was used stored memory in
the AstGen arena, which may get freed after Sema. Now it properly sets
up a new anonymous Decl for error sets and uses a normal decl_val
instruction.
The other usage of `const` ZIR instruction was float literals. These are
now changed to use `float` ZIR instruction when the value fits inside
`zir.Inst.Data` and `float128` otherwise.
AstGen + Sema: implement int_to_enum and enum_to_int. No tests yet; I expect to
have to make some fixes before they will pass tests. Will do that in the
branch before merging.
AstGen: fix struct astgen incorrectly counting decls as fields.
Type/Value: give up on trying to exhaustively list every tag all the
time. This makes the file more manageable. Also found a bug with
i128/u128 this way, since the name of the function was more obvious when
looking at the tag values.
Type: implement abiAlignment and abiSize for structs. This will need to
get more sophisticated at some point, but for now it is progress.
Value: add new `enum_field_index` tag.
Value: add hash_u32, needed when using ArrayHashMap.
2021-04-06 17:43:56 -07:00
|
|
|
|
|
|
|
|
|
2021-04-27 18:36:12 -07:00
|
|
|
|
|
|
|
|
|
2023-05-25 19:23:01 -04:00
|
|
|
|
2020-09-13 19:17:58 -07:00
|
|
|
|
|
|
|
|
|
2023-05-29 11:32:12 -04:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-05-28 10:33:59 -04:00
|
|
|
|
|
|
|
|
|
2023-05-02 15:01:45 -07:00
|
|
|
|
2021-04-26 20:41:07 -07:00
|
|
|
|
2023-05-02 15:01:45 -07:00
|
|
|
|
2021-04-26 20:41:07 -07:00
|
|
|
|
|
|
|
|
|
2023-05-26 16:04:53 -04:00
|
|
|
|
2021-05-05 13:16:14 -07:00
|
|
|
|
2023-08-24 20:43:43 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-05-05 13:16:14 -07:00
|
|
|
|
|
|
|
|
|
2023-05-26 16:04:53 -04:00
|
|
|
|
2021-05-07 14:18:14 -07:00
|
|
|
|
2023-08-21 14:27:34 -07:00
|
|
|
|
2021-05-11 14:17:52 -07:00
|
|
|
|
2023-05-21 04:29:34 -04:00
|
|
|
|
|
|
|
|
|
2021-05-07 14:18:14 -07:00
|
|
|
|
|
|
|
|
|
rework generic function calls
Abridged summary:
* Move `Module.Fn` into `InternPool`.
* Delete a lot of confusing and problematic `Sema` logic related to
generic function calls.
This commit removes `Module.Fn` and replaces it with two new
`InternPool.Tag` values:
* `func_decl` - corresponding to a function declared in the source
code. This one contains line/column numbers, zir_body_inst, etc.
* `func_instance` - one for each monomorphization of a generic
function. Contains a reference to the `func_decl` from whence the
instantiation came, along with the `comptime` parameter values (or
types in the case of `anytype`)
Since `InternPool` provides deduplication on these values, these fields
are now deleted from `Module`:
* `monomorphed_func_keys`
* `monomorphed_funcs`
* `align_stack_fns`
Instead of these, Sema logic for generic function instantiation now
unconditionally evaluates the function prototype expression for every
generic callsite. This is technically required in order for type
coercions to work. The previous code had some dubious, probably wrong
hacks to make things work, such as `hashUncoerced`. I'm not 100% sure
how we were able to eliminate that function and still pass all the
behavior tests, but I'm pretty sure things were still broken without
doing type coercion for every generic function call argument.
After the function prototype is evaluated, it produces a deduplicated
`func_instance` `InternPool.Index` which can then be used for the
generic function call.
Some other nice things made by this simplification are the removal of
`comptime_args_fn_inst` and `preallocated_new_func` from `Sema`, and the
messy logic associated with them.
I have not yet been able to measure the perf of this against master
branch. On one hand, it reduces memory usage and pointer chasing of the
most heavily used `InternPool` Tag - function bodies - but on the other
hand, it does evaluate function prototype expressions more than before.
We will soon find out.
2023-07-03 22:09:30 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-05-22 07:58:02 -04:00
|
|
|
|
|
|
|
|
|
2023-07-17 17:47:59 -07:00
|
|
|
|
rework generic function calls
Abridged summary:
* Move `Module.Fn` into `InternPool`.
* Delete a lot of confusing and problematic `Sema` logic related to
generic function calls.
This commit removes `Module.Fn` and replaces it with two new
`InternPool.Tag` values:
* `func_decl` - corresponding to a function declared in the source
code. This one contains line/column numbers, zir_body_inst, etc.
* `func_instance` - one for each monomorphization of a generic
function. Contains a reference to the `func_decl` from whence the
instantiation came, along with the `comptime` parameter values (or
types in the case of `anytype`)
Since `InternPool` provides deduplication on these values, these fields
are now deleted from `Module`:
* `monomorphed_func_keys`
* `monomorphed_funcs`
* `align_stack_fns`
Instead of these, Sema logic for generic function instantiation now
unconditionally evaluates the function prototype expression for every
generic callsite. This is technically required in order for type
coercions to work. The previous code had some dubious, probably wrong
hacks to make things work, such as `hashUncoerced`. I'm not 100% sure
how we were able to eliminate that function and still pass all the
behavior tests, but I'm pretty sure things were still broken without
doing type coercion for every generic function call argument.
After the function prototype is evaluated, it produces a deduplicated
`func_instance` `InternPool.Index` which can then be used for the
generic function call.
Some other nice things made by this simplification are the removal of
`comptime_args_fn_inst` and `preallocated_new_func` from `Sema`, and the
messy logic associated with them.
I have not yet been able to measure the perf of this against master
branch. On one hand, it reduces memory usage and pointer chasing of the
most heavily used `InternPool` Tag - function bodies - but on the other
hand, it does evaluate function prototype expressions more than before.
We will soon find out.
2023-07-03 22:09:30 -07:00
|
|
|
|
|
|
|
|
|
2021-05-05 16:56:24 -07:00
|
|
|
|
|
|
|
|
|
2023-05-26 16:04:53 -04:00
|
|
|
|
2022-02-06 09:14:15 +01:00
|
|
|
|
2023-05-26 16:04:53 -04:00
|
|
|
|
2023-05-22 07:58:02 -04:00
|
|
|
|
2022-02-05 17:41:03 +01:00
|
|
|
|
|
|
|
|
|
2023-05-26 16:04:53 -04:00
|
|
|
|
2022-02-06 09:14:15 +01:00
|
|
|
|
2023-05-26 16:04:53 -04:00
|
|
|
|
2023-05-22 07:58:02 -04:00
|
|
|
|
2021-05-07 18:52:11 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-08-14 13:06:47 +01:00
|
|
|
|
|
|
|
|
|
stage2: move enum types into the InternPool
Unlike unions and structs, enums are actually *encoded* into the
InternPool directly, rather than using the SegmentedList trick. This
results in them being quite compact, and greatly improved the ergonomics
of using enum types throughout the compiler.
It did however require introducing a new concept to the InternPool which
is an "incomplete" item - something that is added to gain a permanent
Index, but which is then mutated in place. This was necessary because
enum tag values and tag types may reference the namespaces created by
the enum itself, which required constructing the namespace, decl, and
calling analyzeDecl on the decl, which required the decl value, which
required the enum type, which required an InternPool index to be
assigned and for it to be meaningful.
The API for updating enums in place turned out to be quite slick and
efficient - the methods directly populate pre-allocated arrays and
return the information necessary to output the same compilation errors
as before.
2023-05-12 00:07:32 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-05-25 05:47:25 -04:00
|
|
|
|
2023-05-10 12:16:24 -07:00
|
|
|
|
|
|
|
|
|
2023-08-21 14:27:34 -07:00
|
|
|
|
stage2: move enum types into the InternPool
Unlike unions and structs, enums are actually *encoded* into the
InternPool directly, rather than using the SegmentedList trick. This
results in them being quite compact, and greatly improved the ergonomics
of using enum types throughout the compiler.
It did however require introducing a new concept to the InternPool which
is an "incomplete" item - something that is added to gain a permanent
Index, but which is then mutated in place. This was necessary because
enum tag values and tag types may reference the namespaces created by
the enum itself, which required constructing the namespace, decl, and
calling analyzeDecl on the decl, which required the decl value, which
required the enum type, which required an InternPool index to be
assigned and for it to be meaningful.
The API for updating enums in place turned out to be quite slick and
efficient - the methods directly populate pre-allocated arrays and
return the information necessary to output the same compilation errors
as before.
2023-05-12 00:07:32 -07:00
|
|
|
|
2023-05-10 12:16:24 -07:00
|
|
|
|
|
|
|
|
|
stage2: move enum types into the InternPool
Unlike unions and structs, enums are actually *encoded* into the
InternPool directly, rather than using the SegmentedList trick. This
results in them being quite compact, and greatly improved the ergonomics
of using enum types throughout the compiler.
It did however require introducing a new concept to the InternPool which
is an "incomplete" item - something that is added to gain a permanent
Index, but which is then mutated in place. This was necessary because
enum tag values and tag types may reference the namespaces created by
the enum itself, which required constructing the namespace, decl, and
calling analyzeDecl on the decl, which required the decl value, which
required the enum type, which required an InternPool index to be
assigned and for it to be meaningful.
The API for updating enums in place turned out to be quite slick and
efficient - the methods directly populate pre-allocated arrays and
return the information necessary to output the same compilation errors
as before.
2023-05-12 00:07:32 -07:00
|
|
|
|
2023-05-08 16:52:59 -07:00
|
|
|
|
|
|
|
|
|
2023-08-14 13:06:47 +01:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-05-26 16:04:53 -04:00
|
|
|
|
|
|
|
|
|
2021-05-07 18:52:11 -07:00
|
|
|
|
|
|
|
|
|
2023-08-14 13:06:47 +01:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
stage2: *WIP*: rework ZIR memory layout; overhaul source locations
The memory layout for ZIR instructions is completely reworked. See
zir.zig for those changes. Some new types:
* `zir.Code`: a "finished" set of ZIR instructions. Instead of allocating
each instruction independently, there is now a Tag and 8 bytes of
data available for all ZIR instructions. Small instructions fit
within these 8 bytes; larger ones use 4 bytes for an index into
`extra`. There is also `string_bytes` so that we can have 4 byte
references to strings. `zir.Inst.Tag` describes how to interpret
those 8 bytes of data.
- This is shared by all `Block` scopes.
* `Module.WipZirCode`: represents an in-progress `zir.Code`. In this
structure, the arrays are mutable, and get resized as we add/delete
things. There is extra state to keep track of things. This struct is
stored on the stack. Once it is finished, it produces an immutable
`zir.Code`, which will remain on the heap for the duration of a
function's existence.
- This is shared by all `GenZir` scopes.
* `Sema`: represents in-progress semantic analysis of a `zir.Code`.
This data is stored on the stack and is shared among all `Block`
scopes. It is now the main "self" argument to everything in the file
that was previously named `zir_sema.zig`.
Additionally, I moved some logic that was in `Module` into here.
`Module.Fn` now stores its parameter names inside the `zir.Code`,
instead of inside ZIR instructions. When the TZIR memory layout
reworking time comes, codegen will be able to reference this data
directly instead of duplicating it.
astgen.zig is (so far) almost entirely untouched, but nearly all of it
will need to be reworked to adhere to this new memory layout structure.
I have no benchmarks to report yet, as I am still working through
compile errors and fixing various things that I broke in this branch.
Overhaul of Source Locations:
Previously we used `usize` everywhere to mean byte offset, but sometimes
also mean other stuff. This was error prone and also made us do
unnecessary work, and store unnecessary bytes in memory.
Now there are more types involved into source locations, and more ways
to describe a source location.
* AllErrors.Message: embrace the assumption that files always have less
than 2 << 32 bytes.
* SrcLoc gets more complicated, to model more complicated source
locations.
* Introduce LazySrcLoc, which can model interesting source locations
with very little stored state. Useful for avoiding doing unnecessary
work when no compile errors occur.
Also, previously, we had `src: usize` on every ZIR instruction. This is
no longer the case. Each instruction now determines whether it even cares
about source location, and if so, how that source location is stored.
This requires more careful work inside `Sema`, but it results in fewer
bytes stored on the heap, without compromising accuracy and power of
compile error messages.
Miscellaneous:
* std.zig: string literals have more helpful result values for
reporting errors. There is now a lower level API and a higher level
API.
- side note: I noticed that the string literal logic needs some love.
There is some unnecessarily hacky code there.
* cut & pasted some TZIR logic that was in zir.zig to ir.zig. This
probably broke stuff and needs to get fixed.
* Removed type/Enum.zig, type/Union.zig, and type/Struct.zig. I don't
think this quite how this code will be organized. Need some more
careful planning about how to implement structs, unions, enums. They
need to be independent Decls, just like a top level function.
2021-03-15 23:38:38 -07:00
|
|
|
|
|
|
|
|
|
compiler: eliminate Decl.value_arena and Sema.perm_arena
The main motivation for this commit is eliminating Decl.value_arena.
Everything else is dominoes.
Decl.name used to be stored in the GPA, now it is stored in InternPool.
It ended up being simpler to migrate other strings to be interned as
well, such as struct field names, union field names, and a few others.
This ended up requiring a big diff, sorry about that. But the changes
are pretty nice, we finally start to take advantage of InternPool's
existence.
global_error_set and error_name_list are simplified. Now it is a single
ArrayHashMap(NullTerminatedString, void) and the index is the error tag
value.
Module.tmp_hack_arena is re-introduced (it was removed in
eeff407941560ce8eb5b737b2436dfa93cfd3a0c) in order to deal with
comptime_args, optimized_order, and struct and union fields. After
structs and unions get moved into InternPool properly, tmp_hack_arena
can be deleted again.
2023-06-01 23:46:04 -07:00
|
|
|
|
stage2: *WIP*: rework ZIR memory layout; overhaul source locations
The memory layout for ZIR instructions is completely reworked. See
zir.zig for those changes. Some new types:
* `zir.Code`: a "finished" set of ZIR instructions. Instead of allocating
each instruction independently, there is now a Tag and 8 bytes of
data available for all ZIR instructions. Small instructions fit
within these 8 bytes; larger ones use 4 bytes for an index into
`extra`. There is also `string_bytes` so that we can have 4 byte
references to strings. `zir.Inst.Tag` describes how to interpret
those 8 bytes of data.
- This is shared by all `Block` scopes.
* `Module.WipZirCode`: represents an in-progress `zir.Code`. In this
structure, the arrays are mutable, and get resized as we add/delete
things. There is extra state to keep track of things. This struct is
stored on the stack. Once it is finished, it produces an immutable
`zir.Code`, which will remain on the heap for the duration of a
function's existence.
- This is shared by all `GenZir` scopes.
* `Sema`: represents in-progress semantic analysis of a `zir.Code`.
This data is stored on the stack and is shared among all `Block`
scopes. It is now the main "self" argument to everything in the file
that was previously named `zir_sema.zig`.
Additionally, I moved some logic that was in `Module` into here.
`Module.Fn` now stores its parameter names inside the `zir.Code`,
instead of inside ZIR instructions. When the TZIR memory layout
reworking time comes, codegen will be able to reference this data
directly instead of duplicating it.
astgen.zig is (so far) almost entirely untouched, but nearly all of it
will need to be reworked to adhere to this new memory layout structure.
I have no benchmarks to report yet, as I am still working through
compile errors and fixing various things that I broke in this branch.
Overhaul of Source Locations:
Previously we used `usize` everywhere to mean byte offset, but sometimes
also mean other stuff. This was error prone and also made us do
unnecessary work, and store unnecessary bytes in memory.
Now there are more types involved into source locations, and more ways
to describe a source location.
* AllErrors.Message: embrace the assumption that files always have less
than 2 << 32 bytes.
* SrcLoc gets more complicated, to model more complicated source
locations.
* Introduce LazySrcLoc, which can model interesting source locations
with very little stored state. Useful for avoiding doing unnecessary
work when no compile errors occur.
Also, previously, we had `src: usize` on every ZIR instruction. This is
no longer the case. Each instruction now determines whether it even cares
about source location, and if so, how that source location is stored.
This requires more careful work inside `Sema`, but it results in fewer
bytes stored on the heap, without compromising accuracy and power of
compile error messages.
Miscellaneous:
* std.zig: string literals have more helpful result values for
reporting errors. There is now a lower level API and a higher level
API.
- side note: I noticed that the string literal logic needs some love.
There is some unnecessarily hacky code there.
* cut & pasted some TZIR logic that was in zir.zig to ir.zig. This
probably broke stuff and needs to get fixed.
* Removed type/Enum.zig, type/Union.zig, and type/Struct.zig. I don't
think this quite how this code will be organized. Need some more
careful planning about how to implement structs, unions, enums. They
need to be independent Decls, just like a top level function.
2021-03-15 23:38:38 -07:00
|
|
|
|
2020-09-13 19:17:58 -07:00
|
|
|
|
|
|
|
|
|
2023-06-15 13:14:16 +06:00
|
|
|
|
stage2: *WIP*: rework ZIR memory layout; overhaul source locations
The memory layout for ZIR instructions is completely reworked. See
zir.zig for those changes. Some new types:
* `zir.Code`: a "finished" set of ZIR instructions. Instead of allocating
each instruction independently, there is now a Tag and 8 bytes of
data available for all ZIR instructions. Small instructions fit
within these 8 bytes; larger ones use 4 bytes for an index into
`extra`. There is also `string_bytes` so that we can have 4 byte
references to strings. `zir.Inst.Tag` describes how to interpret
those 8 bytes of data.
- This is shared by all `Block` scopes.
* `Module.WipZirCode`: represents an in-progress `zir.Code`. In this
structure, the arrays are mutable, and get resized as we add/delete
things. There is extra state to keep track of things. This struct is
stored on the stack. Once it is finished, it produces an immutable
`zir.Code`, which will remain on the heap for the duration of a
function's existence.
- This is shared by all `GenZir` scopes.
* `Sema`: represents in-progress semantic analysis of a `zir.Code`.
This data is stored on the stack and is shared among all `Block`
scopes. It is now the main "self" argument to everything in the file
that was previously named `zir_sema.zig`.
Additionally, I moved some logic that was in `Module` into here.
`Module.Fn` now stores its parameter names inside the `zir.Code`,
instead of inside ZIR instructions. When the TZIR memory layout
reworking time comes, codegen will be able to reference this data
directly instead of duplicating it.
astgen.zig is (so far) almost entirely untouched, but nearly all of it
will need to be reworked to adhere to this new memory layout structure.
I have no benchmarks to report yet, as I am still working through
compile errors and fixing various things that I broke in this branch.
Overhaul of Source Locations:
Previously we used `usize` everywhere to mean byte offset, but sometimes
also mean other stuff. This was error prone and also made us do
unnecessary work, and store unnecessary bytes in memory.
Now there are more types involved into source locations, and more ways
to describe a source location.
* AllErrors.Message: embrace the assumption that files always have less
than 2 << 32 bytes.
* SrcLoc gets more complicated, to model more complicated source
locations.
* Introduce LazySrcLoc, which can model interesting source locations
with very little stored state. Useful for avoiding doing unnecessary
work when no compile errors occur.
Also, previously, we had `src: usize` on every ZIR instruction. This is
no longer the case. Each instruction now determines whether it even cares
about source location, and if so, how that source location is stored.
This requires more careful work inside `Sema`, but it results in fewer
bytes stored on the heap, without compromising accuracy and power of
compile error messages.
Miscellaneous:
* std.zig: string literals have more helpful result values for
reporting errors. There is now a lower level API and a higher level
API.
- side note: I noticed that the string literal logic needs some love.
There is some unnecessarily hacky code there.
* cut & pasted some TZIR logic that was in zir.zig to ir.zig. This
probably broke stuff and needs to get fixed.
* Removed type/Enum.zig, type/Union.zig, and type/Struct.zig. I don't
think this quite how this code will be organized. Need some more
careful planning about how to implement structs, unions, enums. They
need to be independent Decls, just like a top level function.
2021-03-15 23:38:38 -07:00
|
|
|
|
2020-09-13 19:17:58 -07:00
|
|
|
|
2021-04-27 18:36:12 -07:00
|
|
|
|
|
|
|
|
|
2020-09-13 19:17:58 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-05-08 16:52:59 -07:00
|
|
|
|
|
|
|
|
|
2021-01-01 19:24:02 -07:00
|
|
|
|
|
|
|
|
|
2022-04-19 21:51:08 -07:00
|
|
|
|
2021-06-03 15:39:26 -05:00
|
|
|
|
2020-09-13 19:17:58 -07:00
|
|
|
|
|
|
|
|
|
2022-04-19 21:51:08 -07:00
|
|
|
|
2021-06-03 15:39:26 -05:00
|
|
|
|
2020-09-13 19:17:58 -07:00
|
|
|
|
2021-10-16 12:26:06 -07:00
|
|
|
|
2023-07-18 17:04:28 -04:00
|
|
|
|
2021-10-16 12:26:06 -07:00
|
|
|
|
2023-05-25 05:47:25 -04:00
|
|
|
|
2023-07-18 17:04:28 -04:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-10-16 12:26:06 -07:00
|
|
|
|
|
|
|
|
|
2021-10-22 17:12:12 -07:00
|
|
|
|
2023-07-18 17:04:28 -04:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-08-24 20:43:43 -07:00
|
|
|
|
2021-10-22 17:12:12 -07:00
|
|
|
|
2023-08-24 20:43:43 -07:00
|
|
|
|
|
|
|
|
|
2021-10-22 17:12:12 -07:00
|
|
|
|
2020-09-13 19:17:58 -07:00
|
|
|
|
|
|
|
|
|
2021-01-05 17:33:31 -07:00
|
|
|
|
|
|
|
|
|
2021-03-25 23:00:38 -07:00
|
|
|
|
2021-01-05 17:33:31 -07:00
|
|
|
|
|
|
|
|
|
2022-03-25 22:53:58 +08:00
|
|
|
|
2022-04-19 21:51:08 -07:00
|
|
|
|
|
|
|
|
|
compiler: eliminate Decl.value_arena and Sema.perm_arena
The main motivation for this commit is eliminating Decl.value_arena.
Everything else is dominoes.
Decl.name used to be stored in the GPA, now it is stored in InternPool.
It ended up being simpler to migrate other strings to be interned as
well, such as struct field names, union field names, and a few others.
This ended up requiring a big diff, sorry about that. But the changes
are pretty nice, we finally start to take advantage of InternPool's
existence.
global_error_set and error_name_list are simplified. Now it is a single
ArrayHashMap(NullTerminatedString, void) and the index is the error tag
value.
Module.tmp_hack_arena is re-introduced (it was removed in
eeff407941560ce8eb5b737b2436dfa93cfd3a0c) in order to deal with
comptime_args, optimized_order, and struct and union fields. After
structs and unions get moved into InternPool properly, tmp_hack_arena
can be deleted again.
2023-06-01 23:46:04 -07:00
|
|
|
|
2022-03-25 22:53:58 +08:00
|
|
|
|
2023-06-15 13:14:16 +06:00
|
|
|
|
2022-03-25 22:53:58 +08:00
|
|
|
|
|
|
|
|
|
compiler: eliminate Decl.value_arena and Sema.perm_arena
The main motivation for this commit is eliminating Decl.value_arena.
Everything else is dominoes.
Decl.name used to be stored in the GPA, now it is stored in InternPool.
It ended up being simpler to migrate other strings to be interned as
well, such as struct field names, union field names, and a few others.
This ended up requiring a big diff, sorry about that. But the changes
are pretty nice, we finally start to take advantage of InternPool's
existence.
global_error_set and error_name_list are simplified. Now it is a single
ArrayHashMap(NullTerminatedString, void) and the index is the error tag
value.
Module.tmp_hack_arena is re-introduced (it was removed in
eeff407941560ce8eb5b737b2436dfa93cfd3a0c) in order to deal with
comptime_args, optimized_order, and struct and union fields. After
structs and unions get moved into InternPool properly, tmp_hack_arena
can be deleted again.
2023-06-01 23:46:04 -07:00
|
|
|
|
2022-03-25 22:53:58 +08:00
|
|
|
|
2022-04-19 21:51:08 -07:00
|
|
|
|
compiler: eliminate Decl.value_arena and Sema.perm_arena
The main motivation for this commit is eliminating Decl.value_arena.
Everything else is dominoes.
Decl.name used to be stored in the GPA, now it is stored in InternPool.
It ended up being simpler to migrate other strings to be interned as
well, such as struct field names, union field names, and a few others.
This ended up requiring a big diff, sorry about that. But the changes
are pretty nice, we finally start to take advantage of InternPool's
existence.
global_error_set and error_name_list are simplified. Now it is a single
ArrayHashMap(NullTerminatedString, void) and the index is the error tag
value.
Module.tmp_hack_arena is re-introduced (it was removed in
eeff407941560ce8eb5b737b2436dfa93cfd3a0c) in order to deal with
comptime_args, optimized_order, and struct and union fields. After
structs and unions get moved into InternPool properly, tmp_hack_arena
can be deleted again.
2023-06-01 23:46:04 -07:00
|
|
|
|
2022-03-25 22:53:58 +08:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-10-01 17:45:13 -05:00
|
|
|
|
|
|
|
|
|
2023-05-08 16:52:59 -07:00
|
|
|
|
2021-10-01 17:45:13 -05:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-07-06 15:27:47 -07:00
|
|
|
|
|
|
|
|
|
2022-04-19 21:51:08 -07:00
|
|
|
|
2020-09-13 19:17:58 -07:00
|
|
|
|
2021-10-01 17:45:13 -05:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2022-04-19 21:51:08 -07:00
|
|
|
|
2021-10-01 17:45:13 -05:00
|
|
|
|
2023-05-08 16:52:59 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-06-22 18:46:56 +01:00
|
|
|
|
2023-05-08 16:52:59 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-06-22 18:46:56 +01:00
|
|
|
|
2023-05-08 16:52:59 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-06-22 18:46:56 +01:00
|
|
|
|
2023-05-08 16:52:59 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2022-03-25 22:53:58 +08:00
|
|
|
|
2022-04-19 21:51:08 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-06-15 13:14:16 +06:00
|
|
|
|
2022-03-25 22:53:58 +08:00
|
|
|
|
|
|
|
|
|
2022-04-19 21:51:08 -07:00
|
|
|
|
2022-03-25 22:53:58 +08:00
|
|
|
|
2022-04-19 21:51:08 -07:00
|
|
|
|
|
|
|
|
|
compiler: eliminate Decl.value_arena and Sema.perm_arena
The main motivation for this commit is eliminating Decl.value_arena.
Everything else is dominoes.
Decl.name used to be stored in the GPA, now it is stored in InternPool.
It ended up being simpler to migrate other strings to be interned as
well, such as struct field names, union field names, and a few others.
This ended up requiring a big diff, sorry about that. But the changes
are pretty nice, we finally start to take advantage of InternPool's
existence.
global_error_set and error_name_list are simplified. Now it is a single
ArrayHashMap(NullTerminatedString, void) and the index is the error tag
value.
Module.tmp_hack_arena is re-introduced (it was removed in
eeff407941560ce8eb5b737b2436dfa93cfd3a0c) in order to deal with
comptime_args, optimized_order, and struct and union fields. After
structs and unions get moved into InternPool properly, tmp_hack_arena
can be deleted again.
2023-06-01 23:46:04 -07:00
|
|
|
|
2022-03-25 22:53:58 +08:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-10-01 17:45:13 -05:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2020-09-13 19:17:58 -07:00
|
|
|
|
|
|
|
|
|
2021-10-01 17:45:13 -05:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2022-04-19 21:51:08 -07:00
|
|
|
|
|
|
|
|
|
2021-10-01 17:45:13 -05:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-12-05 12:34:50 +01:00
|
|
|
|
2021-01-16 22:51:01 -07:00
|
|
|
|
|
|
|
|
|
2021-10-01 17:45:13 -05:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2022-04-19 21:51:08 -07:00
|
|
|
|
2021-10-01 17:45:13 -05:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2022-03-25 22:53:58 +08:00
|
|
|
|
2021-10-01 17:45:13 -05:00
|
|
|
|
2022-04-19 21:51:08 -07:00
|
|
|
|
2021-10-01 17:45:13 -05:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-12-05 12:34:50 +01:00
|
|
|
|
2021-09-23 12:17:06 -05:00
|
|
|
|
|
|
|
|
|
2021-10-01 17:45:13 -05:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2022-04-19 21:51:08 -07:00
|
|
|
|
2023-06-12 01:44:12 -04:00
|
|
|
|
2021-10-01 17:45:13 -05:00
|
|
|
|
|
|
|
|
|
2023-05-08 16:52:59 -07:00
|
|
|
|
2023-06-12 01:44:12 -04:00
|
|
|
|
|
|
|
|
|
2021-10-01 17:45:13 -05:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-06-12 01:44:12 -04:00
|
|
|
|
2020-09-13 19:17:58 -07:00
|
|
|
|
|
|
|
|
|
2021-10-01 17:45:13 -05:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2022-04-19 21:51:08 -07:00
|
|
|
|
2023-06-12 01:44:12 -04:00
|
|
|
|
2021-10-01 17:45:13 -05:00
|
|
|
|
|
|
|
|
|
2023-06-12 01:44:12 -04:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-10-01 17:45:13 -05:00
|
|
|
|
2023-06-12 01:44:12 -04:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2020-09-13 19:17:58 -07:00
|
|
|
|
|
|
|
|
|
2023-05-08 16:52:59 -07:00
|
|
|
|
|
|
|
|
|
2021-10-01 17:45:13 -05:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2022-04-19 21:51:08 -07:00
|
|
|
|
|
|
|
|
|
2021-10-01 17:45:13 -05:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-12-30 20:49:02 -07:00
|
|
|
|
2021-10-01 17:45:13 -05:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-10-06 17:05:58 -07:00
|
|
|
|
|
|
|
|
|
2022-11-27 14:17:55 +00:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-10-01 17:45:13 -05:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2022-11-27 01:07:35 -07:00
|
|
|
|
2021-10-01 17:45:13 -05:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2022-11-27 01:07:35 -07:00
|
|
|
|
2021-10-01 17:45:13 -05:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2022-11-27 14:17:55 +00:00
|
|
|
|
2023-02-18 04:10:50 +00:00
|
|
|
|
2022-11-27 14:17:55 +00:00
|
|
|
|
|
|
|
|
|
2023-10-06 21:29:08 -07:00
|
|
|
|
|
|
|
|
|
2022-11-27 14:17:55 +00:00
|
|
|
|
|
|
|
|
|
2021-10-29 00:37:25 +01:00
|
|
|
|
2021-10-01 17:45:13 -05:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-10-29 00:37:25 +01:00
|
|
|
|
2021-10-01 17:45:13 -05:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-01-28 22:38:25 +02:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-10-29 00:37:25 +01:00
|
|
|
|
2021-10-01 17:45:13 -05:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-04-30 11:07:31 -07:00
|
|
|
|
2021-10-01 17:45:13 -05:00
|
|
|
|
2021-04-30 11:07:31 -07:00
|
|
|
|
2021-10-29 00:37:25 +01:00
|
|
|
|
2021-10-01 17:45:13 -05:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2020-09-13 19:17:58 -07:00
|
|
|
|
2021-10-01 17:45:13 -05:00
|
|
|
|
2020-09-13 19:17:58 -07:00
|
|
|
|
2021-10-01 17:45:13 -05:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2022-11-27 14:17:55 +00:00
|
|
|
|
2022-04-19 21:51:08 -07:00
|
|
|
|
|
|
|
|
|
2020-09-13 19:17:58 -07:00
|
|
|
|
2021-10-01 17:45:13 -05:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
stage2: implement simple enums
A simple enum is an enum which has an automatic integer tag type,
all tag values automatically assigned, and no top level declarations.
Such enums are created directly in AstGen and shared by all the
generic/comptime instantiations of the surrounding ZIR code. This
commit implements, but does not yet add any test cases for, simple enums.
A full enum is an enum for which any of the above conditions are not
true. Full enums are created in Sema, and therefore will create a unique
type per generic/comptime instantiation. This commit does not implement
full enums. However the `enum_decl_nonexhaustive` ZIR instruction is
added and the respective Type functions are filled out.
This commit makes an improvement to ZIR code, removing the decls array
and removing the decl_map from AstGen. Instead, decl_ref and
decl_val ZIR instructions index into the `owner_decl.dependencies`
ArrayHashMap. We already need this dependencies array for incremental
compilation purposes, and so repurposing it to also use it for ZIR decl
indexes makes for efficient memory usage.
Similarly, this commit fixes up incorrect memory management by removing
the `const` ZIR instruction. The two places it was used stored memory in
the AstGen arena, which may get freed after Sema. Now it properly sets
up a new anonymous Decl for error sets and uses a normal decl_val
instruction.
The other usage of `const` ZIR instruction was float literals. These are
now changed to use `float` ZIR instruction when the value fits inside
`zir.Inst.Data` and `float128` otherwise.
AstGen + Sema: implement int_to_enum and enum_to_int. No tests yet; I expect to
have to make some fixes before they will pass tests. Will do that in the
branch before merging.
AstGen: fix struct astgen incorrectly counting decls as fields.
Type/Value: give up on trying to exhaustively list every tag all the
time. This makes the file more manageable. Also found a bug with
i128/u128 this way, since the name of the function was more obvious when
looking at the tag values.
Type: implement abiAlignment and abiSize for structs. This will need to
get more sophisticated at some point, but for now it is progress.
Value: add new `enum_field_index` tag.
Value: add hash_u32, needed when using ArrayHashMap.
2021-04-06 17:43:56 -07:00
|
|
|
|
2021-10-01 17:45:13 -05:00
|
|
|
|
|
|
|
|
|
stage2: entry point via std lib and proper updated file detection
Instead of Module setting up the root_scope with the root source file,
instead, Module relies on the package table graph being set up properly,
and inside `update()`, it does the equivalent of `_ = @import("std");`.
This, in term, imports start.zig, which has the logic to call main (or
not). `Module` no longer has `root_scope` - the root source file is no
longer special, it's just in the package table mapped to "root".
I also went ahead and implemented proper detection of updated files.
mtime, inode, size, and source hash are kept in `Scope.File`.
During an update, iterate over `import_table` and stat each file to find
out which ones are updated.
The source hash is redundant with the source hash used by the struct
decl that corresponds to the file, so it should be removed in a future
commit before merging the branch.
* AstGen: add "previously declared here" notes for variables shadowing
decls.
* Parse imports as structs. Module now calls `AstGen.structDeclInner`,
which is called by `AstGen.containerDecl`.
- `importFile` is a bit kludgy with how it handles the top level Decl
that kinda gets merged into the struct decl at the end of the
function. Be on the look out for bugs related to that as well as
possibly cleaner ways to implement this.
* Module: factor out lookupDeclName into lookupIdentifier and lookupNa
* Rename `Scope.Container` to `Scope.Namespace`.
* Delete some dead code.
This branch won't work until `usingnamespace` is implemented because it
relies on `@import("builtin").OutputMode` and `OutputMode` comes from a
`usingnamespace`.
2021-04-09 23:17:50 -07:00
|
|
|
|
2021-12-30 20:49:02 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-09-30 21:43:30 -05:00
|
|
|
|
2021-10-01 17:45:13 -05:00
|
|
|
|
|
|
|
|
|
2023-10-06 17:05:58 -07:00
|
|
|
|
2021-10-01 17:45:13 -05:00
|
|
|
|
2020-09-13 19:17:58 -07:00
|
|
|
|
2021-10-01 17:45:13 -05:00
|
|
|
|
2021-05-06 17:20:45 -07:00
|
|
|
|
2021-10-01 17:45:13 -05:00
|
|
|
|
|
|
|
|
|
2021-05-11 17:34:13 -07:00
|
|
|
|
2023-06-22 18:46:56 +01:00
|
|
|
|
2021-10-01 17:45:13 -05:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2020-09-13 19:17:58 -07:00
|
|
|
|
2021-10-01 17:45:13 -05:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
stage2: entry point via std lib and proper updated file detection
Instead of Module setting up the root_scope with the root source file,
instead, Module relies on the package table graph being set up properly,
and inside `update()`, it does the equivalent of `_ = @import("std");`.
This, in term, imports start.zig, which has the logic to call main (or
not). `Module` no longer has `root_scope` - the root source file is no
longer special, it's just in the package table mapped to "root".
I also went ahead and implemented proper detection of updated files.
mtime, inode, size, and source hash are kept in `Scope.File`.
During an update, iterate over `import_table` and stat each file to find
out which ones are updated.
The source hash is redundant with the source hash used by the struct
decl that corresponds to the file, so it should be removed in a future
commit before merging the branch.
* AstGen: add "previously declared here" notes for variables shadowing
decls.
* Parse imports as structs. Module now calls `AstGen.structDeclInner`,
which is called by `AstGen.containerDecl`.
- `importFile` is a bit kludgy with how it handles the top level Decl
that kinda gets merged into the struct decl at the end of the
function. Be on the look out for bugs related to that as well as
possibly cleaner ways to implement this.
* Module: factor out lookupDeclName into lookupIdentifier and lookupNa
* Rename `Scope.Container` to `Scope.Namespace`.
* Delete some dead code.
This branch won't work until `usingnamespace` is implemented because it
relies on `@import("builtin").OutputMode` and `OutputMode` comes from a
`usingnamespace`.
2021-04-09 23:17:50 -07:00
|
|
|
|
2021-10-01 17:45:13 -05:00
|
|
|
|
|
|
|
|
|
2021-12-30 20:49:02 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-10-01 17:45:13 -05:00
|
|
|
|
stage2: entry point via std lib and proper updated file detection
Instead of Module setting up the root_scope with the root source file,
instead, Module relies on the package table graph being set up properly,
and inside `update()`, it does the equivalent of `_ = @import("std");`.
This, in term, imports start.zig, which has the logic to call main (or
not). `Module` no longer has `root_scope` - the root source file is no
longer special, it's just in the package table mapped to "root".
I also went ahead and implemented proper detection of updated files.
mtime, inode, size, and source hash are kept in `Scope.File`.
During an update, iterate over `import_table` and stat each file to find
out which ones are updated.
The source hash is redundant with the source hash used by the struct
decl that corresponds to the file, so it should be removed in a future
commit before merging the branch.
* AstGen: add "previously declared here" notes for variables shadowing
decls.
* Parse imports as structs. Module now calls `AstGen.structDeclInner`,
which is called by `AstGen.containerDecl`.
- `importFile` is a bit kludgy with how it handles the top level Decl
that kinda gets merged into the struct decl at the end of the
function. Be on the look out for bugs related to that as well as
possibly cleaner ways to implement this.
* Module: factor out lookupDeclName into lookupIdentifier and lookupNa
* Rename `Scope.Container` to `Scope.Namespace`.
* Delete some dead code.
This branch won't work until `usingnamespace` is implemented because it
relies on `@import("builtin").OutputMode` and `OutputMode` comes from a
`usingnamespace`.
2021-04-09 23:17:50 -07:00
|
|
|
|
2021-10-29 00:37:25 +01:00
|
|
|
|
2021-10-01 17:45:13 -05:00
|
|
|
|
2021-04-14 11:26:53 -07:00
|
|
|
|
2021-10-01 17:45:13 -05:00
|
|
|
|
2023-02-02 17:32:06 -07:00
|
|
|
|
2021-10-01 17:45:13 -05:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2020-09-13 19:17:58 -07:00
|
|
|
|
2021-10-01 17:45:13 -05:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
stage2: entry point via std lib and proper updated file detection
Instead of Module setting up the root_scope with the root source file,
instead, Module relies on the package table graph being set up properly,
and inside `update()`, it does the equivalent of `_ = @import("std");`.
This, in term, imports start.zig, which has the logic to call main (or
not). `Module` no longer has `root_scope` - the root source file is no
longer special, it's just in the package table mapped to "root".
I also went ahead and implemented proper detection of updated files.
mtime, inode, size, and source hash are kept in `Scope.File`.
During an update, iterate over `import_table` and stat each file to find
out which ones are updated.
The source hash is redundant with the source hash used by the struct
decl that corresponds to the file, so it should be removed in a future
commit before merging the branch.
* AstGen: add "previously declared here" notes for variables shadowing
decls.
* Parse imports as structs. Module now calls `AstGen.structDeclInner`,
which is called by `AstGen.containerDecl`.
- `importFile` is a bit kludgy with how it handles the top level Decl
that kinda gets merged into the struct decl at the end of the
function. Be on the look out for bugs related to that as well as
possibly cleaner ways to implement this.
* Module: factor out lookupDeclName into lookupIdentifier and lookupNa
* Rename `Scope.Container` to `Scope.Namespace`.
* Delete some dead code.
This branch won't work until `usingnamespace` is implemented because it
relies on `@import("builtin").OutputMode` and `OutputMode` comes from a
`usingnamespace`.
2021-04-09 23:17:50 -07:00
|
|
|
|
2021-10-01 17:45:13 -05:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
stage2 generics improvements: anytype and param type exprs
AstGen result locations now have a `coerced_ty` tag which is the same as
`ty` except it assumes that Sema will do a coercion, so it does not
redundantly add an `as` instruction into the ZIR code. This results in
cleaner ZIR and about a 14% reduction of ZIR bytes.
param and param_comptime ZIR instructions now have a block body for
their type expressions. This allows Sema to skip evaluation of the
block in the case that the parameter is comptime-provided. It also
allows a new mechanism to function: when evaluating type expressions of
generic functions, if it would depend on another parameter, it returns
`error.GenericPoison` which bubbles up and then is caught by the
param/param_comptime instruction and then handled.
This allows parameters to be evaluated independently so that the type
info for functions which have comptime or anytype parameters will still
have types populated for parameters that do not depend on values of
previous parameters (because evaluation of their param blocks will return
successfully instead of `error.GenericPoison`).
It also makes iteration over the block that contains function parameters
slightly more efficient since it now only contains the param
instructions.
Finally, it fixes the case where a generic function type expression contains
a function prototype. Formerly, this situation would cause shared state
to clobber each other; now it is in a proper tree structure so that
can't happen. This fix also required adding a field to Sema
`comptime_args_fn_inst` to make sure that the `comptime_args` field
passed into Sema is applied to the correct `func` instruction.
Source location for `node_offset_asm_ret_ty` is fixed; it was pointing at
the asm output name rather than the return type as intended.
Generic function instantiation is fixed, notably with respect to
parameter type expressions that depend on previous parameters, and with
respect to types which must be always comptime-known. This involves
passing all the comptime arguments at a callsite of a generic function,
and allowing the generic function semantic analysis to coerce the values
to the proper types (since it has access to the evaluated parameter type
expressions) and then decide based on the type whether the parameter is
runtime known or not. In the case of explicitly marked `comptime`
parameters, there is a check at the semantic analysis of the `call`
instruction.
Semantic analysis of `call` instructions does type coercion on the
arguments, which is needed both for generic functions and to make up for
using `coerced_ty` result locations (mentioned above).
Tasks left in this branch:
* Implement the memoization table.
* Add test coverage.
* Improve error reporting and source locations for compile errors.
2021-08-04 21:11:31 -07:00
|
|
|
|
2021-10-01 17:45:13 -05:00
|
|
|
|
stage2 generics improvements: anytype and param type exprs
AstGen result locations now have a `coerced_ty` tag which is the same as
`ty` except it assumes that Sema will do a coercion, so it does not
redundantly add an `as` instruction into the ZIR code. This results in
cleaner ZIR and about a 14% reduction of ZIR bytes.
param and param_comptime ZIR instructions now have a block body for
their type expressions. This allows Sema to skip evaluation of the
block in the case that the parameter is comptime-provided. It also
allows a new mechanism to function: when evaluating type expressions of
generic functions, if it would depend on another parameter, it returns
`error.GenericPoison` which bubbles up and then is caught by the
param/param_comptime instruction and then handled.
This allows parameters to be evaluated independently so that the type
info for functions which have comptime or anytype parameters will still
have types populated for parameters that do not depend on values of
previous parameters (because evaluation of their param blocks will return
successfully instead of `error.GenericPoison`).
It also makes iteration over the block that contains function parameters
slightly more efficient since it now only contains the param
instructions.
Finally, it fixes the case where a generic function type expression contains
a function prototype. Formerly, this situation would cause shared state
to clobber each other; now it is in a proper tree structure so that
can't happen. This fix also required adding a field to Sema
`comptime_args_fn_inst` to make sure that the `comptime_args` field
passed into Sema is applied to the correct `func` instruction.
Source location for `node_offset_asm_ret_ty` is fixed; it was pointing at
the asm output name rather than the return type as intended.
Generic function instantiation is fixed, notably with respect to
parameter type expressions that depend on previous parameters, and with
respect to types which must be always comptime-known. This involves
passing all the comptime arguments at a callsite of a generic function,
and allowing the generic function semantic analysis to coerce the values
to the proper types (since it has access to the evaluated parameter type
expressions) and then decide based on the type whether the parameter is
runtime known or not. In the case of explicitly marked `comptime`
parameters, there is a check at the semantic analysis of the `call`
instruction.
Semantic analysis of `call` instructions does type coercion on the
arguments, which is needed both for generic functions and to make up for
using `coerced_ty` result locations (mentioned above).
Tasks left in this branch:
* Implement the memoization table.
* Add test coverage.
* Improve error reporting and source locations for compile errors.
2021-08-04 21:11:31 -07:00
|
|
|
|
2021-10-01 17:45:13 -05:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-01-02 14:28:03 -07:00
|
|
|
|
2021-10-01 17:45:13 -05:00
|
|
|
|
2021-01-02 12:32:30 -07:00
|
|
|
|
compiler: eliminate Decl.value_arena and Sema.perm_arena
The main motivation for this commit is eliminating Decl.value_arena.
Everything else is dominoes.
Decl.name used to be stored in the GPA, now it is stored in InternPool.
It ended up being simpler to migrate other strings to be interned as
well, such as struct field names, union field names, and a few others.
This ended up requiring a big diff, sorry about that. But the changes
are pretty nice, we finally start to take advantage of InternPool's
existence.
global_error_set and error_name_list are simplified. Now it is a single
ArrayHashMap(NullTerminatedString, void) and the index is the error tag
value.
Module.tmp_hack_arena is re-introduced (it was removed in
eeff407941560ce8eb5b737b2436dfa93cfd3a0c) in order to deal with
comptime_args, optimized_order, and struct and union fields. After
structs and unions get moved into InternPool properly, tmp_hack_arena
can be deleted again.
2023-06-01 23:46:04 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-10-01 17:45:13 -05:00
|
|
|
|
|
|
|
|
|
2021-10-29 00:37:25 +01:00
|
|
|
|
2023-10-06 17:05:58 -07:00
|
|
|
|
2021-10-01 17:45:13 -05:00
|
|
|
|
|
|
|
|
|
2022-01-08 04:29:49 +01:00
|
|
|
|
2023-10-06 17:05:58 -07:00
|
|
|
|
2022-01-08 04:29:49 +01:00
|
|
|
|
|
|
|
|
|
2021-10-01 17:45:13 -05:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-01-02 14:28:03 -07:00
|
|
|
|
2021-10-01 17:45:13 -05:00
|
|
|
|
2022-11-27 14:17:55 +00:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-02-17 01:44:08 +00:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2022-11-27 14:17:55 +00:00
|
|
|
|
|
|
|
|
|
2023-10-06 21:29:08 -07:00
|
|
|
|
2022-11-27 14:17:55 +00:00
|
|
|
|
|
|
|
|
|
2023-10-06 21:29:08 -07:00
|
|
|
|
2022-11-27 14:17:55 +00:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-02-18 04:10:50 +00:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-06-15 13:14:16 +06:00
|
|
|
|
2022-11-27 14:17:55 +00:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2020-09-13 19:17:58 -07:00
|
|
|
|
|
|
|
|
|
2021-10-17 18:57:54 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-12-30 20:49:02 -07:00
|
|
|
|
2021-10-17 18:57:54 -07:00
|
|
|
|
2023-10-06 21:29:08 -07:00
|
|
|
|
2021-10-17 18:57:54 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2022-04-19 21:51:08 -07:00
|
|
|
|
2021-11-24 16:58:01 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-10-17 18:57:54 -07:00
|
|
|
|
|
|
|
|
|
2021-01-16 22:51:01 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-10-01 17:45:13 -05:00
|
|
|
|
|
|
|
|
|
2021-01-16 22:51:01 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2022-09-09 18:55:58 +03:00
|
|
|
|
2023-08-26 19:42:56 -04:00
|
|
|
|
2022-09-09 18:55:58 +03:00
|
|
|
|
|
|
|
|
|
2023-08-26 19:42:56 -04:00
|
|
|
|
2022-09-09 18:55:58 +03:00
|
|
|
|
|
|
|
|
|
2021-01-16 22:51:01 -07:00
|
|
|
|
|
|
|
|
|
2021-10-29 00:37:25 +01:00
|
|
|
|
2021-01-16 22:51:01 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
stage2: *WIP*: rework ZIR memory layout; overhaul source locations
The memory layout for ZIR instructions is completely reworked. See
zir.zig for those changes. Some new types:
* `zir.Code`: a "finished" set of ZIR instructions. Instead of allocating
each instruction independently, there is now a Tag and 8 bytes of
data available for all ZIR instructions. Small instructions fit
within these 8 bytes; larger ones use 4 bytes for an index into
`extra`. There is also `string_bytes` so that we can have 4 byte
references to strings. `zir.Inst.Tag` describes how to interpret
those 8 bytes of data.
- This is shared by all `Block` scopes.
* `Module.WipZirCode`: represents an in-progress `zir.Code`. In this
structure, the arrays are mutable, and get resized as we add/delete
things. There is extra state to keep track of things. This struct is
stored on the stack. Once it is finished, it produces an immutable
`zir.Code`, which will remain on the heap for the duration of a
function's existence.
- This is shared by all `GenZir` scopes.
* `Sema`: represents in-progress semantic analysis of a `zir.Code`.
This data is stored on the stack and is shared among all `Block`
scopes. It is now the main "self" argument to everything in the file
that was previously named `zir_sema.zig`.
Additionally, I moved some logic that was in `Module` into here.
`Module.Fn` now stores its parameter names inside the `zir.Code`,
instead of inside ZIR instructions. When the TZIR memory layout
reworking time comes, codegen will be able to reference this data
directly instead of duplicating it.
astgen.zig is (so far) almost entirely untouched, but nearly all of it
will need to be reworked to adhere to this new memory layout structure.
I have no benchmarks to report yet, as I am still working through
compile errors and fixing various things that I broke in this branch.
Overhaul of Source Locations:
Previously we used `usize` everywhere to mean byte offset, but sometimes
also mean other stuff. This was error prone and also made us do
unnecessary work, and store unnecessary bytes in memory.
Now there are more types involved into source locations, and more ways
to describe a source location.
* AllErrors.Message: embrace the assumption that files always have less
than 2 << 32 bytes.
* SrcLoc gets more complicated, to model more complicated source
locations.
* Introduce LazySrcLoc, which can model interesting source locations
with very little stored state. Useful for avoiding doing unnecessary
work when no compile errors occur.
Also, previously, we had `src: usize` on every ZIR instruction. This is
no longer the case. Each instruction now determines whether it even cares
about source location, and if so, how that source location is stored.
This requires more careful work inside `Sema`, but it results in fewer
bytes stored on the heap, without compromising accuracy and power of
compile error messages.
Miscellaneous:
* std.zig: string literals have more helpful result values for
reporting errors. There is now a lower level API and a higher level
API.
- side note: I noticed that the string literal logic needs some love.
There is some unnecessarily hacky code there.
* cut & pasted some TZIR logic that was in zir.zig to ir.zig. This
probably broke stuff and needs to get fixed.
* Removed type/Enum.zig, type/Union.zig, and type/Struct.zig. I don't
think this quite how this code will be organized. Need some more
careful planning about how to implement structs, unions, enums. They
need to be independent Decls, just like a top level function.
2021-03-15 23:38:38 -07:00
|
|
|
|
|
|
|
|
|
2023-05-02 15:01:45 -07:00
|
|
|
|
stage2: *WIP*: rework ZIR memory layout; overhaul source locations
The memory layout for ZIR instructions is completely reworked. See
zir.zig for those changes. Some new types:
* `zir.Code`: a "finished" set of ZIR instructions. Instead of allocating
each instruction independently, there is now a Tag and 8 bytes of
data available for all ZIR instructions. Small instructions fit
within these 8 bytes; larger ones use 4 bytes for an index into
`extra`. There is also `string_bytes` so that we can have 4 byte
references to strings. `zir.Inst.Tag` describes how to interpret
those 8 bytes of data.
- This is shared by all `Block` scopes.
* `Module.WipZirCode`: represents an in-progress `zir.Code`. In this
structure, the arrays are mutable, and get resized as we add/delete
things. There is extra state to keep track of things. This struct is
stored on the stack. Once it is finished, it produces an immutable
`zir.Code`, which will remain on the heap for the duration of a
function's existence.
- This is shared by all `GenZir` scopes.
* `Sema`: represents in-progress semantic analysis of a `zir.Code`.
This data is stored on the stack and is shared among all `Block`
scopes. It is now the main "self" argument to everything in the file
that was previously named `zir_sema.zig`.
Additionally, I moved some logic that was in `Module` into here.
`Module.Fn` now stores its parameter names inside the `zir.Code`,
instead of inside ZIR instructions. When the TZIR memory layout
reworking time comes, codegen will be able to reference this data
directly instead of duplicating it.
astgen.zig is (so far) almost entirely untouched, but nearly all of it
will need to be reworked to adhere to this new memory layout structure.
I have no benchmarks to report yet, as I am still working through
compile errors and fixing various things that I broke in this branch.
Overhaul of Source Locations:
Previously we used `usize` everywhere to mean byte offset, but sometimes
also mean other stuff. This was error prone and also made us do
unnecessary work, and store unnecessary bytes in memory.
Now there are more types involved into source locations, and more ways
to describe a source location.
* AllErrors.Message: embrace the assumption that files always have less
than 2 << 32 bytes.
* SrcLoc gets more complicated, to model more complicated source
locations.
* Introduce LazySrcLoc, which can model interesting source locations
with very little stored state. Useful for avoiding doing unnecessary
work when no compile errors occur.
Also, previously, we had `src: usize` on every ZIR instruction. This is
no longer the case. Each instruction now determines whether it even cares
about source location, and if so, how that source location is stored.
This requires more careful work inside `Sema`, but it results in fewer
bytes stored on the heap, without compromising accuracy and power of
compile error messages.
Miscellaneous:
* std.zig: string literals have more helpful result values for
reporting errors. There is now a lower level API and a higher level
API.
- side note: I noticed that the string literal logic needs some love.
There is some unnecessarily hacky code there.
* cut & pasted some TZIR logic that was in zir.zig to ir.zig. This
probably broke stuff and needs to get fixed.
* Removed type/Enum.zig, type/Union.zig, and type/Struct.zig. I don't
think this quite how this code will be organized. Need some more
careful planning about how to implement structs, unions, enums. They
need to be independent Decls, just like a top level function.
2021-03-15 23:38:38 -07:00
|
|
|
|
2021-01-16 22:51:01 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-10-29 00:37:25 +01:00
|
|
|
|
stage2: *WIP*: rework ZIR memory layout; overhaul source locations
The memory layout for ZIR instructions is completely reworked. See
zir.zig for those changes. Some new types:
* `zir.Code`: a "finished" set of ZIR instructions. Instead of allocating
each instruction independently, there is now a Tag and 8 bytes of
data available for all ZIR instructions. Small instructions fit
within these 8 bytes; larger ones use 4 bytes for an index into
`extra`. There is also `string_bytes` so that we can have 4 byte
references to strings. `zir.Inst.Tag` describes how to interpret
those 8 bytes of data.
- This is shared by all `Block` scopes.
* `Module.WipZirCode`: represents an in-progress `zir.Code`. In this
structure, the arrays are mutable, and get resized as we add/delete
things. There is extra state to keep track of things. This struct is
stored on the stack. Once it is finished, it produces an immutable
`zir.Code`, which will remain on the heap for the duration of a
function's existence.
- This is shared by all `GenZir` scopes.
* `Sema`: represents in-progress semantic analysis of a `zir.Code`.
This data is stored on the stack and is shared among all `Block`
scopes. It is now the main "self" argument to everything in the file
that was previously named `zir_sema.zig`.
Additionally, I moved some logic that was in `Module` into here.
`Module.Fn` now stores its parameter names inside the `zir.Code`,
instead of inside ZIR instructions. When the TZIR memory layout
reworking time comes, codegen will be able to reference this data
directly instead of duplicating it.
astgen.zig is (so far) almost entirely untouched, but nearly all of it
will need to be reworked to adhere to this new memory layout structure.
I have no benchmarks to report yet, as I am still working through
compile errors and fixing various things that I broke in this branch.
Overhaul of Source Locations:
Previously we used `usize` everywhere to mean byte offset, but sometimes
also mean other stuff. This was error prone and also made us do
unnecessary work, and store unnecessary bytes in memory.
Now there are more types involved into source locations, and more ways
to describe a source location.
* AllErrors.Message: embrace the assumption that files always have less
than 2 << 32 bytes.
* SrcLoc gets more complicated, to model more complicated source
locations.
* Introduce LazySrcLoc, which can model interesting source locations
with very little stored state. Useful for avoiding doing unnecessary
work when no compile errors occur.
Also, previously, we had `src: usize` on every ZIR instruction. This is
no longer the case. Each instruction now determines whether it even cares
about source location, and if so, how that source location is stored.
This requires more careful work inside `Sema`, but it results in fewer
bytes stored on the heap, without compromising accuracy and power of
compile error messages.
Miscellaneous:
* std.zig: string literals have more helpful result values for
reporting errors. There is now a lower level API and a higher level
API.
- side note: I noticed that the string literal logic needs some love.
There is some unnecessarily hacky code there.
* cut & pasted some TZIR logic that was in zir.zig to ir.zig. This
probably broke stuff and needs to get fixed.
* Removed type/Enum.zig, type/Union.zig, and type/Struct.zig. I don't
think this quite how this code will be organized. Need some more
careful planning about how to implement structs, unions, enums. They
need to be independent Decls, just like a top level function.
2021-03-15 23:38:38 -07:00
|
|
|
|
|
|
|
|
|
2021-01-16 22:51:01 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-10-29 00:37:25 +01:00
|
|
|
|
2021-01-16 22:51:01 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-10-29 00:37:25 +01:00
|
|
|
|
stage2: *WIP*: rework ZIR memory layout; overhaul source locations
The memory layout for ZIR instructions is completely reworked. See
zir.zig for those changes. Some new types:
* `zir.Code`: a "finished" set of ZIR instructions. Instead of allocating
each instruction independently, there is now a Tag and 8 bytes of
data available for all ZIR instructions. Small instructions fit
within these 8 bytes; larger ones use 4 bytes for an index into
`extra`. There is also `string_bytes` so that we can have 4 byte
references to strings. `zir.Inst.Tag` describes how to interpret
those 8 bytes of data.
- This is shared by all `Block` scopes.
* `Module.WipZirCode`: represents an in-progress `zir.Code`. In this
structure, the arrays are mutable, and get resized as we add/delete
things. There is extra state to keep track of things. This struct is
stored on the stack. Once it is finished, it produces an immutable
`zir.Code`, which will remain on the heap for the duration of a
function's existence.
- This is shared by all `GenZir` scopes.
* `Sema`: represents in-progress semantic analysis of a `zir.Code`.
This data is stored on the stack and is shared among all `Block`
scopes. It is now the main "self" argument to everything in the file
that was previously named `zir_sema.zig`.
Additionally, I moved some logic that was in `Module` into here.
`Module.Fn` now stores its parameter names inside the `zir.Code`,
instead of inside ZIR instructions. When the TZIR memory layout
reworking time comes, codegen will be able to reference this data
directly instead of duplicating it.
astgen.zig is (so far) almost entirely untouched, but nearly all of it
will need to be reworked to adhere to this new memory layout structure.
I have no benchmarks to report yet, as I am still working through
compile errors and fixing various things that I broke in this branch.
Overhaul of Source Locations:
Previously we used `usize` everywhere to mean byte offset, but sometimes
also mean other stuff. This was error prone and also made us do
unnecessary work, and store unnecessary bytes in memory.
Now there are more types involved into source locations, and more ways
to describe a source location.
* AllErrors.Message: embrace the assumption that files always have less
than 2 << 32 bytes.
* SrcLoc gets more complicated, to model more complicated source
locations.
* Introduce LazySrcLoc, which can model interesting source locations
with very little stored state. Useful for avoiding doing unnecessary
work when no compile errors occur.
Also, previously, we had `src: usize` on every ZIR instruction. This is
no longer the case. Each instruction now determines whether it even cares
about source location, and if so, how that source location is stored.
This requires more careful work inside `Sema`, but it results in fewer
bytes stored on the heap, without compromising accuracy and power of
compile error messages.
Miscellaneous:
* std.zig: string literals have more helpful result values for
reporting errors. There is now a lower level API and a higher level
API.
- side note: I noticed that the string literal logic needs some love.
There is some unnecessarily hacky code there.
* cut & pasted some TZIR logic that was in zir.zig to ir.zig. This
probably broke stuff and needs to get fixed.
* Removed type/Enum.zig, type/Union.zig, and type/Struct.zig. I don't
think this quite how this code will be organized. Need some more
careful planning about how to implement structs, unions, enums. They
need to be independent Decls, just like a top level function.
2021-03-15 23:38:38 -07:00
|
|
|
|
2021-01-16 22:51:01 -07:00
|
|
|
|
|
|
|
|
|
stage2: *WIP*: rework ZIR memory layout; overhaul source locations
The memory layout for ZIR instructions is completely reworked. See
zir.zig for those changes. Some new types:
* `zir.Code`: a "finished" set of ZIR instructions. Instead of allocating
each instruction independently, there is now a Tag and 8 bytes of
data available for all ZIR instructions. Small instructions fit
within these 8 bytes; larger ones use 4 bytes for an index into
`extra`. There is also `string_bytes` so that we can have 4 byte
references to strings. `zir.Inst.Tag` describes how to interpret
those 8 bytes of data.
- This is shared by all `Block` scopes.
* `Module.WipZirCode`: represents an in-progress `zir.Code`. In this
structure, the arrays are mutable, and get resized as we add/delete
things. There is extra state to keep track of things. This struct is
stored on the stack. Once it is finished, it produces an immutable
`zir.Code`, which will remain on the heap for the duration of a
function's existence.
- This is shared by all `GenZir` scopes.
* `Sema`: represents in-progress semantic analysis of a `zir.Code`.
This data is stored on the stack and is shared among all `Block`
scopes. It is now the main "self" argument to everything in the file
that was previously named `zir_sema.zig`.
Additionally, I moved some logic that was in `Module` into here.
`Module.Fn` now stores its parameter names inside the `zir.Code`,
instead of inside ZIR instructions. When the TZIR memory layout
reworking time comes, codegen will be able to reference this data
directly instead of duplicating it.
astgen.zig is (so far) almost entirely untouched, but nearly all of it
will need to be reworked to adhere to this new memory layout structure.
I have no benchmarks to report yet, as I am still working through
compile errors and fixing various things that I broke in this branch.
Overhaul of Source Locations:
Previously we used `usize` everywhere to mean byte offset, but sometimes
also mean other stuff. This was error prone and also made us do
unnecessary work, and store unnecessary bytes in memory.
Now there are more types involved into source locations, and more ways
to describe a source location.
* AllErrors.Message: embrace the assumption that files always have less
than 2 << 32 bytes.
* SrcLoc gets more complicated, to model more complicated source
locations.
* Introduce LazySrcLoc, which can model interesting source locations
with very little stored state. Useful for avoiding doing unnecessary
work when no compile errors occur.
Also, previously, we had `src: usize` on every ZIR instruction. This is
no longer the case. Each instruction now determines whether it even cares
about source location, and if so, how that source location is stored.
This requires more careful work inside `Sema`, but it results in fewer
bytes stored on the heap, without compromising accuracy and power of
compile error messages.
Miscellaneous:
* std.zig: string literals have more helpful result values for
reporting errors. There is now a lower level API and a higher level
API.
- side note: I noticed that the string literal logic needs some love.
There is some unnecessarily hacky code there.
* cut & pasted some TZIR logic that was in zir.zig to ir.zig. This
probably broke stuff and needs to get fixed.
* Removed type/Enum.zig, type/Union.zig, and type/Struct.zig. I don't
think this quite how this code will be organized. Need some more
careful planning about how to implement structs, unions, enums. They
need to be independent Decls, just like a top level function.
2021-03-15 23:38:38 -07:00
|
|
|
|
|
|
|
|
|
2022-09-09 18:55:58 +03:00
|
|
|
|
stage2: *WIP*: rework ZIR memory layout; overhaul source locations
The memory layout for ZIR instructions is completely reworked. See
zir.zig for those changes. Some new types:
* `zir.Code`: a "finished" set of ZIR instructions. Instead of allocating
each instruction independently, there is now a Tag and 8 bytes of
data available for all ZIR instructions. Small instructions fit
within these 8 bytes; larger ones use 4 bytes for an index into
`extra`. There is also `string_bytes` so that we can have 4 byte
references to strings. `zir.Inst.Tag` describes how to interpret
those 8 bytes of data.
- This is shared by all `Block` scopes.
* `Module.WipZirCode`: represents an in-progress `zir.Code`. In this
structure, the arrays are mutable, and get resized as we add/delete
things. There is extra state to keep track of things. This struct is
stored on the stack. Once it is finished, it produces an immutable
`zir.Code`, which will remain on the heap for the duration of a
function's existence.
- This is shared by all `GenZir` scopes.
* `Sema`: represents in-progress semantic analysis of a `zir.Code`.
This data is stored on the stack and is shared among all `Block`
scopes. It is now the main "self" argument to everything in the file
that was previously named `zir_sema.zig`.
Additionally, I moved some logic that was in `Module` into here.
`Module.Fn` now stores its parameter names inside the `zir.Code`,
instead of inside ZIR instructions. When the TZIR memory layout
reworking time comes, codegen will be able to reference this data
directly instead of duplicating it.
astgen.zig is (so far) almost entirely untouched, but nearly all of it
will need to be reworked to adhere to this new memory layout structure.
I have no benchmarks to report yet, as I am still working through
compile errors and fixing various things that I broke in this branch.
Overhaul of Source Locations:
Previously we used `usize` everywhere to mean byte offset, but sometimes
also mean other stuff. This was error prone and also made us do
unnecessary work, and store unnecessary bytes in memory.
Now there are more types involved into source locations, and more ways
to describe a source location.
* AllErrors.Message: embrace the assumption that files always have less
than 2 << 32 bytes.
* SrcLoc gets more complicated, to model more complicated source
locations.
* Introduce LazySrcLoc, which can model interesting source locations
with very little stored state. Useful for avoiding doing unnecessary
work when no compile errors occur.
Also, previously, we had `src: usize` on every ZIR instruction. This is
no longer the case. Each instruction now determines whether it even cares
about source location, and if so, how that source location is stored.
This requires more careful work inside `Sema`, but it results in fewer
bytes stored on the heap, without compromising accuracy and power of
compile error messages.
Miscellaneous:
* std.zig: string literals have more helpful result values for
reporting errors. There is now a lower level API and a higher level
API.
- side note: I noticed that the string literal logic needs some love.
There is some unnecessarily hacky code there.
* cut & pasted some TZIR logic that was in zir.zig to ir.zig. This
probably broke stuff and needs to get fixed.
* Removed type/Enum.zig, type/Union.zig, and type/Struct.zig. I don't
think this quite how this code will be organized. Need some more
careful planning about how to implement structs, unions, enums. They
need to be independent Decls, just like a top level function.
2021-03-15 23:38:38 -07:00
|
|
|
|
2021-01-16 22:51:01 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-10-01 17:45:13 -05:00
|
|
|
|
2021-04-16 14:44:02 -07:00
|
|
|
|
2021-08-30 19:22:04 -07:00
|
|
|
|
2021-04-16 14:44:02 -07:00
|
|
|
|
stage2: *WIP*: rework ZIR memory layout; overhaul source locations
The memory layout for ZIR instructions is completely reworked. See
zir.zig for those changes. Some new types:
* `zir.Code`: a "finished" set of ZIR instructions. Instead of allocating
each instruction independently, there is now a Tag and 8 bytes of
data available for all ZIR instructions. Small instructions fit
within these 8 bytes; larger ones use 4 bytes for an index into
`extra`. There is also `string_bytes` so that we can have 4 byte
references to strings. `zir.Inst.Tag` describes how to interpret
those 8 bytes of data.
- This is shared by all `Block` scopes.
* `Module.WipZirCode`: represents an in-progress `zir.Code`. In this
structure, the arrays are mutable, and get resized as we add/delete
things. There is extra state to keep track of things. This struct is
stored on the stack. Once it is finished, it produces an immutable
`zir.Code`, which will remain on the heap for the duration of a
function's existence.
- This is shared by all `GenZir` scopes.
* `Sema`: represents in-progress semantic analysis of a `zir.Code`.
This data is stored on the stack and is shared among all `Block`
scopes. It is now the main "self" argument to everything in the file
that was previously named `zir_sema.zig`.
Additionally, I moved some logic that was in `Module` into here.
`Module.Fn` now stores its parameter names inside the `zir.Code`,
instead of inside ZIR instructions. When the TZIR memory layout
reworking time comes, codegen will be able to reference this data
directly instead of duplicating it.
astgen.zig is (so far) almost entirely untouched, but nearly all of it
will need to be reworked to adhere to this new memory layout structure.
I have no benchmarks to report yet, as I am still working through
compile errors and fixing various things that I broke in this branch.
Overhaul of Source Locations:
Previously we used `usize` everywhere to mean byte offset, but sometimes
also mean other stuff. This was error prone and also made us do
unnecessary work, and store unnecessary bytes in memory.
Now there are more types involved into source locations, and more ways
to describe a source location.
* AllErrors.Message: embrace the assumption that files always have less
than 2 << 32 bytes.
* SrcLoc gets more complicated, to model more complicated source
locations.
* Introduce LazySrcLoc, which can model interesting source locations
with very little stored state. Useful for avoiding doing unnecessary
work when no compile errors occur.
Also, previously, we had `src: usize` on every ZIR instruction. This is
no longer the case. Each instruction now determines whether it even cares
about source location, and if so, how that source location is stored.
This requires more careful work inside `Sema`, but it results in fewer
bytes stored on the heap, without compromising accuracy and power of
compile error messages.
Miscellaneous:
* std.zig: string literals have more helpful result values for
reporting errors. There is now a lower level API and a higher level
API.
- side note: I noticed that the string literal logic needs some love.
There is some unnecessarily hacky code there.
* cut & pasted some TZIR logic that was in zir.zig to ir.zig. This
probably broke stuff and needs to get fixed.
* Removed type/Enum.zig, type/Union.zig, and type/Struct.zig. I don't
think this quite how this code will be organized. Need some more
careful planning about how to implement structs, unions, enums. They
need to be independent Decls, just like a top level function.
2021-03-15 23:38:38 -07:00
|
|
|
|
|
|
|
|
|
2021-08-30 19:22:04 -07:00
|
|
|
|
2021-04-16 14:44:02 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
stage2: *WIP*: rework ZIR memory layout; overhaul source locations
The memory layout for ZIR instructions is completely reworked. See
zir.zig for those changes. Some new types:
* `zir.Code`: a "finished" set of ZIR instructions. Instead of allocating
each instruction independently, there is now a Tag and 8 bytes of
data available for all ZIR instructions. Small instructions fit
within these 8 bytes; larger ones use 4 bytes for an index into
`extra`. There is also `string_bytes` so that we can have 4 byte
references to strings. `zir.Inst.Tag` describes how to interpret
those 8 bytes of data.
- This is shared by all `Block` scopes.
* `Module.WipZirCode`: represents an in-progress `zir.Code`. In this
structure, the arrays are mutable, and get resized as we add/delete
things. There is extra state to keep track of things. This struct is
stored on the stack. Once it is finished, it produces an immutable
`zir.Code`, which will remain on the heap for the duration of a
function's existence.
- This is shared by all `GenZir` scopes.
* `Sema`: represents in-progress semantic analysis of a `zir.Code`.
This data is stored on the stack and is shared among all `Block`
scopes. It is now the main "self" argument to everything in the file
that was previously named `zir_sema.zig`.
Additionally, I moved some logic that was in `Module` into here.
`Module.Fn` now stores its parameter names inside the `zir.Code`,
instead of inside ZIR instructions. When the TZIR memory layout
reworking time comes, codegen will be able to reference this data
directly instead of duplicating it.
astgen.zig is (so far) almost entirely untouched, but nearly all of it
will need to be reworked to adhere to this new memory layout structure.
I have no benchmarks to report yet, as I am still working through
compile errors and fixing various things that I broke in this branch.
Overhaul of Source Locations:
Previously we used `usize` everywhere to mean byte offset, but sometimes
also mean other stuff. This was error prone and also made us do
unnecessary work, and store unnecessary bytes in memory.
Now there are more types involved into source locations, and more ways
to describe a source location.
* AllErrors.Message: embrace the assumption that files always have less
than 2 << 32 bytes.
* SrcLoc gets more complicated, to model more complicated source
locations.
* Introduce LazySrcLoc, which can model interesting source locations
with very little stored state. Useful for avoiding doing unnecessary
work when no compile errors occur.
Also, previously, we had `src: usize` on every ZIR instruction. This is
no longer the case. Each instruction now determines whether it even cares
about source location, and if so, how that source location is stored.
This requires more careful work inside `Sema`, but it results in fewer
bytes stored on the heap, without compromising accuracy and power of
compile error messages.
Miscellaneous:
* std.zig: string literals have more helpful result values for
reporting errors. There is now a lower level API and a higher level
API.
- side note: I noticed that the string literal logic needs some love.
There is some unnecessarily hacky code there.
* cut & pasted some TZIR logic that was in zir.zig to ir.zig. This
probably broke stuff and needs to get fixed.
* Removed type/Enum.zig, type/Union.zig, and type/Struct.zig. I don't
think this quite how this code will be organized. Need some more
careful planning about how to implement structs, unions, enums. They
need to be independent Decls, just like a top level function.
2021-03-15 23:38:38 -07:00
|
|
|
|
2023-07-18 17:12:19 -07:00
|
|
|
|
|
|
|
|
|
stage2: *WIP*: rework ZIR memory layout; overhaul source locations
The memory layout for ZIR instructions is completely reworked. See
zir.zig for those changes. Some new types:
* `zir.Code`: a "finished" set of ZIR instructions. Instead of allocating
each instruction independently, there is now a Tag and 8 bytes of
data available for all ZIR instructions. Small instructions fit
within these 8 bytes; larger ones use 4 bytes for an index into
`extra`. There is also `string_bytes` so that we can have 4 byte
references to strings. `zir.Inst.Tag` describes how to interpret
those 8 bytes of data.
- This is shared by all `Block` scopes.
* `Module.WipZirCode`: represents an in-progress `zir.Code`. In this
structure, the arrays are mutable, and get resized as we add/delete
things. There is extra state to keep track of things. This struct is
stored on the stack. Once it is finished, it produces an immutable
`zir.Code`, which will remain on the heap for the duration of a
function's existence.
- This is shared by all `GenZir` scopes.
* `Sema`: represents in-progress semantic analysis of a `zir.Code`.
This data is stored on the stack and is shared among all `Block`
scopes. It is now the main "self" argument to everything in the file
that was previously named `zir_sema.zig`.
Additionally, I moved some logic that was in `Module` into here.
`Module.Fn` now stores its parameter names inside the `zir.Code`,
instead of inside ZIR instructions. When the TZIR memory layout
reworking time comes, codegen will be able to reference this data
directly instead of duplicating it.
astgen.zig is (so far) almost entirely untouched, but nearly all of it
will need to be reworked to adhere to this new memory layout structure.
I have no benchmarks to report yet, as I am still working through
compile errors and fixing various things that I broke in this branch.
Overhaul of Source Locations:
Previously we used `usize` everywhere to mean byte offset, but sometimes
also mean other stuff. This was error prone and also made us do
unnecessary work, and store unnecessary bytes in memory.
Now there are more types involved into source locations, and more ways
to describe a source location.
* AllErrors.Message: embrace the assumption that files always have less
than 2 << 32 bytes.
* SrcLoc gets more complicated, to model more complicated source
locations.
* Introduce LazySrcLoc, which can model interesting source locations
with very little stored state. Useful for avoiding doing unnecessary
work when no compile errors occur.
Also, previously, we had `src: usize` on every ZIR instruction. This is
no longer the case. Each instruction now determines whether it even cares
about source location, and if so, how that source location is stored.
This requires more careful work inside `Sema`, but it results in fewer
bytes stored on the heap, without compromising accuracy and power of
compile error messages.
Miscellaneous:
* std.zig: string literals have more helpful result values for
reporting errors. There is now a lower level API and a higher level
API.
- side note: I noticed that the string literal logic needs some love.
There is some unnecessarily hacky code there.
* cut & pasted some TZIR logic that was in zir.zig to ir.zig. This
probably broke stuff and needs to get fixed.
* Removed type/Enum.zig, type/Union.zig, and type/Struct.zig. I don't
think this quite how this code will be organized. Need some more
careful planning about how to implement structs, unions, enums. They
need to be independent Decls, just like a top level function.
2021-03-15 23:38:38 -07:00
|
|
|
|
|
|
|
|
|
2022-07-14 21:26:01 +03:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2022-07-15 11:38:16 +03:00
|
|
|
|
2022-07-14 21:26:01 +03:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
stage2: *WIP*: rework ZIR memory layout; overhaul source locations
The memory layout for ZIR instructions is completely reworked. See
zir.zig for those changes. Some new types:
* `zir.Code`: a "finished" set of ZIR instructions. Instead of allocating
each instruction independently, there is now a Tag and 8 bytes of
data available for all ZIR instructions. Small instructions fit
within these 8 bytes; larger ones use 4 bytes for an index into
`extra`. There is also `string_bytes` so that we can have 4 byte
references to strings. `zir.Inst.Tag` describes how to interpret
those 8 bytes of data.
- This is shared by all `Block` scopes.
* `Module.WipZirCode`: represents an in-progress `zir.Code`. In this
structure, the arrays are mutable, and get resized as we add/delete
things. There is extra state to keep track of things. This struct is
stored on the stack. Once it is finished, it produces an immutable
`zir.Code`, which will remain on the heap for the duration of a
function's existence.
- This is shared by all `GenZir` scopes.
* `Sema`: represents in-progress semantic analysis of a `zir.Code`.
This data is stored on the stack and is shared among all `Block`
scopes. It is now the main "self" argument to everything in the file
that was previously named `zir_sema.zig`.
Additionally, I moved some logic that was in `Module` into here.
`Module.Fn` now stores its parameter names inside the `zir.Code`,
instead of inside ZIR instructions. When the TZIR memory layout
reworking time comes, codegen will be able to reference this data
directly instead of duplicating it.
astgen.zig is (so far) almost entirely untouched, but nearly all of it
will need to be reworked to adhere to this new memory layout structure.
I have no benchmarks to report yet, as I am still working through
compile errors and fixing various things that I broke in this branch.
Overhaul of Source Locations:
Previously we used `usize` everywhere to mean byte offset, but sometimes
also mean other stuff. This was error prone and also made us do
unnecessary work, and store unnecessary bytes in memory.
Now there are more types involved into source locations, and more ways
to describe a source location.
* AllErrors.Message: embrace the assumption that files always have less
than 2 << 32 bytes.
* SrcLoc gets more complicated, to model more complicated source
locations.
* Introduce LazySrcLoc, which can model interesting source locations
with very little stored state. Useful for avoiding doing unnecessary
work when no compile errors occur.
Also, previously, we had `src: usize` on every ZIR instruction. This is
no longer the case. Each instruction now determines whether it even cares
about source location, and if so, how that source location is stored.
This requires more careful work inside `Sema`, but it results in fewer
bytes stored on the heap, without compromising accuracy and power of
compile error messages.
Miscellaneous:
* std.zig: string literals have more helpful result values for
reporting errors. There is now a lower level API and a higher level
API.
- side note: I noticed that the string literal logic needs some love.
There is some unnecessarily hacky code there.
* cut & pasted some TZIR logic that was in zir.zig to ir.zig. This
probably broke stuff and needs to get fixed.
* Removed type/Enum.zig, type/Union.zig, and type/Struct.zig. I don't
think this quite how this code will be organized. Need some more
careful planning about how to implement structs, unions, enums. They
need to be independent Decls, just like a top level function.
2021-03-15 23:38:38 -07:00
|
|
|
|
|
|
|
|
|
2022-07-15 11:38:16 +03:00
|
|
|
|
stage2: *WIP*: rework ZIR memory layout; overhaul source locations
The memory layout for ZIR instructions is completely reworked. See
zir.zig for those changes. Some new types:
* `zir.Code`: a "finished" set of ZIR instructions. Instead of allocating
each instruction independently, there is now a Tag and 8 bytes of
data available for all ZIR instructions. Small instructions fit
within these 8 bytes; larger ones use 4 bytes for an index into
`extra`. There is also `string_bytes` so that we can have 4 byte
references to strings. `zir.Inst.Tag` describes how to interpret
those 8 bytes of data.
- This is shared by all `Block` scopes.
* `Module.WipZirCode`: represents an in-progress `zir.Code`. In this
structure, the arrays are mutable, and get resized as we add/delete
things. There is extra state to keep track of things. This struct is
stored on the stack. Once it is finished, it produces an immutable
`zir.Code`, which will remain on the heap for the duration of a
function's existence.
- This is shared by all `GenZir` scopes.
* `Sema`: represents in-progress semantic analysis of a `zir.Code`.
This data is stored on the stack and is shared among all `Block`
scopes. It is now the main "self" argument to everything in the file
that was previously named `zir_sema.zig`.
Additionally, I moved some logic that was in `Module` into here.
`Module.Fn` now stores its parameter names inside the `zir.Code`,
instead of inside ZIR instructions. When the TZIR memory layout
reworking time comes, codegen will be able to reference this data
directly instead of duplicating it.
astgen.zig is (so far) almost entirely untouched, but nearly all of it
will need to be reworked to adhere to this new memory layout structure.
I have no benchmarks to report yet, as I am still working through
compile errors and fixing various things that I broke in this branch.
Overhaul of Source Locations:
Previously we used `usize` everywhere to mean byte offset, but sometimes
also mean other stuff. This was error prone and also made us do
unnecessary work, and store unnecessary bytes in memory.
Now there are more types involved into source locations, and more ways
to describe a source location.
* AllErrors.Message: embrace the assumption that files always have less
than 2 << 32 bytes.
* SrcLoc gets more complicated, to model more complicated source
locations.
* Introduce LazySrcLoc, which can model interesting source locations
with very little stored state. Useful for avoiding doing unnecessary
work when no compile errors occur.
Also, previously, we had `src: usize` on every ZIR instruction. This is
no longer the case. Each instruction now determines whether it even cares
about source location, and if so, how that source location is stored.
This requires more careful work inside `Sema`, but it results in fewer
bytes stored on the heap, without compromising accuracy and power of
compile error messages.
Miscellaneous:
* std.zig: string literals have more helpful result values for
reporting errors. There is now a lower level API and a higher level
API.
- side note: I noticed that the string literal logic needs some love.
There is some unnecessarily hacky code there.
* cut & pasted some TZIR logic that was in zir.zig to ir.zig. This
probably broke stuff and needs to get fixed.
* Removed type/Enum.zig, type/Union.zig, and type/Struct.zig. I don't
think this quite how this code will be organized. Need some more
careful planning about how to implement structs, unions, enums. They
need to be independent Decls, just like a top level function.
2021-03-15 23:38:38 -07:00
|
|
|
|
2022-07-15 11:38:16 +03:00
|
|
|
|
stage2: *WIP*: rework ZIR memory layout; overhaul source locations
The memory layout for ZIR instructions is completely reworked. See
zir.zig for those changes. Some new types:
* `zir.Code`: a "finished" set of ZIR instructions. Instead of allocating
each instruction independently, there is now a Tag and 8 bytes of
data available for all ZIR instructions. Small instructions fit
within these 8 bytes; larger ones use 4 bytes for an index into
`extra`. There is also `string_bytes` so that we can have 4 byte
references to strings. `zir.Inst.Tag` describes how to interpret
those 8 bytes of data.
- This is shared by all `Block` scopes.
* `Module.WipZirCode`: represents an in-progress `zir.Code`. In this
structure, the arrays are mutable, and get resized as we add/delete
things. There is extra state to keep track of things. This struct is
stored on the stack. Once it is finished, it produces an immutable
`zir.Code`, which will remain on the heap for the duration of a
function's existence.
- This is shared by all `GenZir` scopes.
* `Sema`: represents in-progress semantic analysis of a `zir.Code`.
This data is stored on the stack and is shared among all `Block`
scopes. It is now the main "self" argument to everything in the file
that was previously named `zir_sema.zig`.
Additionally, I moved some logic that was in `Module` into here.
`Module.Fn` now stores its parameter names inside the `zir.Code`,
instead of inside ZIR instructions. When the TZIR memory layout
reworking time comes, codegen will be able to reference this data
directly instead of duplicating it.
astgen.zig is (so far) almost entirely untouched, but nearly all of it
will need to be reworked to adhere to this new memory layout structure.
I have no benchmarks to report yet, as I am still working through
compile errors and fixing various things that I broke in this branch.
Overhaul of Source Locations:
Previously we used `usize` everywhere to mean byte offset, but sometimes
also mean other stuff. This was error prone and also made us do
unnecessary work, and store unnecessary bytes in memory.
Now there are more types involved into source locations, and more ways
to describe a source location.
* AllErrors.Message: embrace the assumption that files always have less
than 2 << 32 bytes.
* SrcLoc gets more complicated, to model more complicated source
locations.
* Introduce LazySrcLoc, which can model interesting source locations
with very little stored state. Useful for avoiding doing unnecessary
work when no compile errors occur.
Also, previously, we had `src: usize` on every ZIR instruction. This is
no longer the case. Each instruction now determines whether it even cares
about source location, and if so, how that source location is stored.
This requires more careful work inside `Sema`, but it results in fewer
bytes stored on the heap, without compromising accuracy and power of
compile error messages.
Miscellaneous:
* std.zig: string literals have more helpful result values for
reporting errors. There is now a lower level API and a higher level
API.
- side note: I noticed that the string literal logic needs some love.
There is some unnecessarily hacky code there.
* cut & pasted some TZIR logic that was in zir.zig to ir.zig. This
probably broke stuff and needs to get fixed.
* Removed type/Enum.zig, type/Union.zig, and type/Struct.zig. I don't
think this quite how this code will be organized. Need some more
careful planning about how to implement structs, unions, enums. They
need to be independent Decls, just like a top level function.
2021-03-15 23:38:38 -07:00
|
|
|
|
|
|
|
|
|
2021-04-28 22:43:26 -07:00
|
|
|
|
2022-07-14 21:26:01 +03:00
|
|
|
|
2023-06-22 18:46:56 +01:00
|
|
|
|
2022-07-15 11:38:16 +03:00
|
|
|
|
stage2: *WIP*: rework ZIR memory layout; overhaul source locations
The memory layout for ZIR instructions is completely reworked. See
zir.zig for those changes. Some new types:
* `zir.Code`: a "finished" set of ZIR instructions. Instead of allocating
each instruction independently, there is now a Tag and 8 bytes of
data available for all ZIR instructions. Small instructions fit
within these 8 bytes; larger ones use 4 bytes for an index into
`extra`. There is also `string_bytes` so that we can have 4 byte
references to strings. `zir.Inst.Tag` describes how to interpret
those 8 bytes of data.
- This is shared by all `Block` scopes.
* `Module.WipZirCode`: represents an in-progress `zir.Code`. In this
structure, the arrays are mutable, and get resized as we add/delete
things. There is extra state to keep track of things. This struct is
stored on the stack. Once it is finished, it produces an immutable
`zir.Code`, which will remain on the heap for the duration of a
function's existence.
- This is shared by all `GenZir` scopes.
* `Sema`: represents in-progress semantic analysis of a `zir.Code`.
This data is stored on the stack and is shared among all `Block`
scopes. It is now the main "self" argument to everything in the file
that was previously named `zir_sema.zig`.
Additionally, I moved some logic that was in `Module` into here.
`Module.Fn` now stores its parameter names inside the `zir.Code`,
instead of inside ZIR instructions. When the TZIR memory layout
reworking time comes, codegen will be able to reference this data
directly instead of duplicating it.
astgen.zig is (so far) almost entirely untouched, but nearly all of it
will need to be reworked to adhere to this new memory layout structure.
I have no benchmarks to report yet, as I am still working through
compile errors and fixing various things that I broke in this branch.
Overhaul of Source Locations:
Previously we used `usize` everywhere to mean byte offset, but sometimes
also mean other stuff. This was error prone and also made us do
unnecessary work, and store unnecessary bytes in memory.
Now there are more types involved into source locations, and more ways
to describe a source location.
* AllErrors.Message: embrace the assumption that files always have less
than 2 << 32 bytes.
* SrcLoc gets more complicated, to model more complicated source
locations.
* Introduce LazySrcLoc, which can model interesting source locations
with very little stored state. Useful for avoiding doing unnecessary
work when no compile errors occur.
Also, previously, we had `src: usize` on every ZIR instruction. This is
no longer the case. Each instruction now determines whether it even cares
about source location, and if so, how that source location is stored.
This requires more careful work inside `Sema`, but it results in fewer
bytes stored on the heap, without compromising accuracy and power of
compile error messages.
Miscellaneous:
* std.zig: string literals have more helpful result values for
reporting errors. There is now a lower level API and a higher level
API.
- side note: I noticed that the string literal logic needs some love.
There is some unnecessarily hacky code there.
* cut & pasted some TZIR logic that was in zir.zig to ir.zig. This
probably broke stuff and needs to get fixed.
* Removed type/Enum.zig, type/Union.zig, and type/Struct.zig. I don't
think this quite how this code will be organized. Need some more
careful planning about how to implement structs, unions, enums. They
need to be independent Decls, just like a top level function.
2021-03-15 23:38:38 -07:00
|
|
|
|
2021-03-25 00:37:52 -07:00
|
|
|
|
2021-04-28 22:43:26 -07:00
|
|
|
|
2022-07-14 21:26:01 +03:00
|
|
|
|
2021-03-17 22:54:56 -07:00
|
|
|
|
stage2: *WIP*: rework ZIR memory layout; overhaul source locations
The memory layout for ZIR instructions is completely reworked. See
zir.zig for those changes. Some new types:
* `zir.Code`: a "finished" set of ZIR instructions. Instead of allocating
each instruction independently, there is now a Tag and 8 bytes of
data available for all ZIR instructions. Small instructions fit
within these 8 bytes; larger ones use 4 bytes for an index into
`extra`. There is also `string_bytes` so that we can have 4 byte
references to strings. `zir.Inst.Tag` describes how to interpret
those 8 bytes of data.
- This is shared by all `Block` scopes.
* `Module.WipZirCode`: represents an in-progress `zir.Code`. In this
structure, the arrays are mutable, and get resized as we add/delete
things. There is extra state to keep track of things. This struct is
stored on the stack. Once it is finished, it produces an immutable
`zir.Code`, which will remain on the heap for the duration of a
function's existence.
- This is shared by all `GenZir` scopes.
* `Sema`: represents in-progress semantic analysis of a `zir.Code`.
This data is stored on the stack and is shared among all `Block`
scopes. It is now the main "self" argument to everything in the file
that was previously named `zir_sema.zig`.
Additionally, I moved some logic that was in `Module` into here.
`Module.Fn` now stores its parameter names inside the `zir.Code`,
instead of inside ZIR instructions. When the TZIR memory layout
reworking time comes, codegen will be able to reference this data
directly instead of duplicating it.
astgen.zig is (so far) almost entirely untouched, but nearly all of it
will need to be reworked to adhere to this new memory layout structure.
I have no benchmarks to report yet, as I am still working through
compile errors and fixing various things that I broke in this branch.
Overhaul of Source Locations:
Previously we used `usize` everywhere to mean byte offset, but sometimes
also mean other stuff. This was error prone and also made us do
unnecessary work, and store unnecessary bytes in memory.
Now there are more types involved into source locations, and more ways
to describe a source location.
* AllErrors.Message: embrace the assumption that files always have less
than 2 << 32 bytes.
* SrcLoc gets more complicated, to model more complicated source
locations.
* Introduce LazySrcLoc, which can model interesting source locations
with very little stored state. Useful for avoiding doing unnecessary
work when no compile errors occur.
Also, previously, we had `src: usize` on every ZIR instruction. This is
no longer the case. Each instruction now determines whether it even cares
about source location, and if so, how that source location is stored.
This requires more careful work inside `Sema`, but it results in fewer
bytes stored on the heap, without compromising accuracy and power of
compile error messages.
Miscellaneous:
* std.zig: string literals have more helpful result values for
reporting errors. There is now a lower level API and a higher level
API.
- side note: I noticed that the string literal logic needs some love.
There is some unnecessarily hacky code there.
* cut & pasted some TZIR logic that was in zir.zig to ir.zig. This
probably broke stuff and needs to get fixed.
* Removed type/Enum.zig, type/Union.zig, and type/Struct.zig. I don't
think this quite how this code will be organized. Need some more
careful planning about how to implement structs, unions, enums. They
need to be independent Decls, just like a top level function.
2021-03-15 23:38:38 -07:00
|
|
|
|
2021-04-28 22:43:26 -07:00
|
|
|
|
2022-07-14 21:26:01 +03:00
|
|
|
|
|
|
|
|
|
2023-06-22 18:46:56 +01:00
|
|
|
|
2022-07-15 11:38:16 +03:00
|
|
|
|
stage2: *WIP*: rework ZIR memory layout; overhaul source locations
The memory layout for ZIR instructions is completely reworked. See
zir.zig for those changes. Some new types:
* `zir.Code`: a "finished" set of ZIR instructions. Instead of allocating
each instruction independently, there is now a Tag and 8 bytes of
data available for all ZIR instructions. Small instructions fit
within these 8 bytes; larger ones use 4 bytes for an index into
`extra`. There is also `string_bytes` so that we can have 4 byte
references to strings. `zir.Inst.Tag` describes how to interpret
those 8 bytes of data.
- This is shared by all `Block` scopes.
* `Module.WipZirCode`: represents an in-progress `zir.Code`. In this
structure, the arrays are mutable, and get resized as we add/delete
things. There is extra state to keep track of things. This struct is
stored on the stack. Once it is finished, it produces an immutable
`zir.Code`, which will remain on the heap for the duration of a
function's existence.
- This is shared by all `GenZir` scopes.
* `Sema`: represents in-progress semantic analysis of a `zir.Code`.
This data is stored on the stack and is shared among all `Block`
scopes. It is now the main "self" argument to everything in the file
that was previously named `zir_sema.zig`.
Additionally, I moved some logic that was in `Module` into here.
`Module.Fn` now stores its parameter names inside the `zir.Code`,
instead of inside ZIR instructions. When the TZIR memory layout
reworking time comes, codegen will be able to reference this data
directly instead of duplicating it.
astgen.zig is (so far) almost entirely untouched, but nearly all of it
will need to be reworked to adhere to this new memory layout structure.
I have no benchmarks to report yet, as I am still working through
compile errors and fixing various things that I broke in this branch.
Overhaul of Source Locations:
Previously we used `usize` everywhere to mean byte offset, but sometimes
also mean other stuff. This was error prone and also made us do
unnecessary work, and store unnecessary bytes in memory.
Now there are more types involved into source locations, and more ways
to describe a source location.
* AllErrors.Message: embrace the assumption that files always have less
than 2 << 32 bytes.
* SrcLoc gets more complicated, to model more complicated source
locations.
* Introduce LazySrcLoc, which can model interesting source locations
with very little stored state. Useful for avoiding doing unnecessary
work when no compile errors occur.
Also, previously, we had `src: usize` on every ZIR instruction. This is
no longer the case. Each instruction now determines whether it even cares
about source location, and if so, how that source location is stored.
This requires more careful work inside `Sema`, but it results in fewer
bytes stored on the heap, without compromising accuracy and power of
compile error messages.
Miscellaneous:
* std.zig: string literals have more helpful result values for
reporting errors. There is now a lower level API and a higher level
API.
- side note: I noticed that the string literal logic needs some love.
There is some unnecessarily hacky code there.
* cut & pasted some TZIR logic that was in zir.zig to ir.zig. This
probably broke stuff and needs to get fixed.
* Removed type/Enum.zig, type/Union.zig, and type/Struct.zig. I don't
think this quite how this code will be organized. Need some more
careful planning about how to implement structs, unions, enums. They
need to be independent Decls, just like a top level function.
2021-03-15 23:38:38 -07:00
|
|
|
|
|
|
|
|
|
2021-04-28 22:43:26 -07:00
|
|
|
|
2021-05-02 15:06:32 -07:00
|
|
|
|
2022-07-14 21:26:01 +03:00
|
|
|
|
2023-06-22 18:46:56 +01:00
|
|
|
|
2022-07-15 11:38:16 +03:00
|
|
|
|
stage2: *WIP*: rework ZIR memory layout; overhaul source locations
The memory layout for ZIR instructions is completely reworked. See
zir.zig for those changes. Some new types:
* `zir.Code`: a "finished" set of ZIR instructions. Instead of allocating
each instruction independently, there is now a Tag and 8 bytes of
data available for all ZIR instructions. Small instructions fit
within these 8 bytes; larger ones use 4 bytes for an index into
`extra`. There is also `string_bytes` so that we can have 4 byte
references to strings. `zir.Inst.Tag` describes how to interpret
those 8 bytes of data.
- This is shared by all `Block` scopes.
* `Module.WipZirCode`: represents an in-progress `zir.Code`. In this
structure, the arrays are mutable, and get resized as we add/delete
things. There is extra state to keep track of things. This struct is
stored on the stack. Once it is finished, it produces an immutable
`zir.Code`, which will remain on the heap for the duration of a
function's existence.
- This is shared by all `GenZir` scopes.
* `Sema`: represents in-progress semantic analysis of a `zir.Code`.
This data is stored on the stack and is shared among all `Block`
scopes. It is now the main "self" argument to everything in the file
that was previously named `zir_sema.zig`.
Additionally, I moved some logic that was in `Module` into here.
`Module.Fn` now stores its parameter names inside the `zir.Code`,
instead of inside ZIR instructions. When the TZIR memory layout
reworking time comes, codegen will be able to reference this data
directly instead of duplicating it.
astgen.zig is (so far) almost entirely untouched, but nearly all of it
will need to be reworked to adhere to this new memory layout structure.
I have no benchmarks to report yet, as I am still working through
compile errors and fixing various things that I broke in this branch.
Overhaul of Source Locations:
Previously we used `usize` everywhere to mean byte offset, but sometimes
also mean other stuff. This was error prone and also made us do
unnecessary work, and store unnecessary bytes in memory.
Now there are more types involved into source locations, and more ways
to describe a source location.
* AllErrors.Message: embrace the assumption that files always have less
than 2 << 32 bytes.
* SrcLoc gets more complicated, to model more complicated source
locations.
* Introduce LazySrcLoc, which can model interesting source locations
with very little stored state. Useful for avoiding doing unnecessary
work when no compile errors occur.
Also, previously, we had `src: usize` on every ZIR instruction. This is
no longer the case. Each instruction now determines whether it even cares
about source location, and if so, how that source location is stored.
This requires more careful work inside `Sema`, but it results in fewer
bytes stored on the heap, without compromising accuracy and power of
compile error messages.
Miscellaneous:
* std.zig: string literals have more helpful result values for
reporting errors. There is now a lower level API and a higher level
API.
- side note: I noticed that the string literal logic needs some love.
There is some unnecessarily hacky code there.
* cut & pasted some TZIR logic that was in zir.zig to ir.zig. This
probably broke stuff and needs to get fixed.
* Removed type/Enum.zig, type/Union.zig, and type/Struct.zig. I don't
think this quite how this code will be organized. Need some more
careful planning about how to implement structs, unions, enums. They
need to be independent Decls, just like a top level function.
2021-03-15 23:38:38 -07:00
|
|
|
|
2022-06-07 17:48:53 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2022-07-14 21:26:01 +03:00
|
|
|
|
2022-06-07 17:48:53 -07:00
|
|
|
|
2022-12-11 20:41:42 +01:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2022-06-07 17:48:53 -07:00
|
|
|
|
2021-04-28 22:43:26 -07:00
|
|
|
|
2021-05-02 15:06:32 -07:00
|
|
|
|
2021-04-28 22:43:26 -07:00
|
|
|
|
2022-07-14 21:26:01 +03:00
|
|
|
|
stage2: *WIP*: rework ZIR memory layout; overhaul source locations
The memory layout for ZIR instructions is completely reworked. See
zir.zig for those changes. Some new types:
* `zir.Code`: a "finished" set of ZIR instructions. Instead of allocating
each instruction independently, there is now a Tag and 8 bytes of
data available for all ZIR instructions. Small instructions fit
within these 8 bytes; larger ones use 4 bytes for an index into
`extra`. There is also `string_bytes` so that we can have 4 byte
references to strings. `zir.Inst.Tag` describes how to interpret
those 8 bytes of data.
- This is shared by all `Block` scopes.
* `Module.WipZirCode`: represents an in-progress `zir.Code`. In this
structure, the arrays are mutable, and get resized as we add/delete
things. There is extra state to keep track of things. This struct is
stored on the stack. Once it is finished, it produces an immutable
`zir.Code`, which will remain on the heap for the duration of a
function's existence.
- This is shared by all `GenZir` scopes.
* `Sema`: represents in-progress semantic analysis of a `zir.Code`.
This data is stored on the stack and is shared among all `Block`
scopes. It is now the main "self" argument to everything in the file
that was previously named `zir_sema.zig`.
Additionally, I moved some logic that was in `Module` into here.
`Module.Fn` now stores its parameter names inside the `zir.Code`,
instead of inside ZIR instructions. When the TZIR memory layout
reworking time comes, codegen will be able to reference this data
directly instead of duplicating it.
astgen.zig is (so far) almost entirely untouched, but nearly all of it
will need to be reworked to adhere to this new memory layout structure.
I have no benchmarks to report yet, as I am still working through
compile errors and fixing various things that I broke in this branch.
Overhaul of Source Locations:
Previously we used `usize` everywhere to mean byte offset, but sometimes
also mean other stuff. This was error prone and also made us do
unnecessary work, and store unnecessary bytes in memory.
Now there are more types involved into source locations, and more ways
to describe a source location.
* AllErrors.Message: embrace the assumption that files always have less
than 2 << 32 bytes.
* SrcLoc gets more complicated, to model more complicated source
locations.
* Introduce LazySrcLoc, which can model interesting source locations
with very little stored state. Useful for avoiding doing unnecessary
work when no compile errors occur.
Also, previously, we had `src: usize` on every ZIR instruction. This is
no longer the case. Each instruction now determines whether it even cares
about source location, and if so, how that source location is stored.
This requires more careful work inside `Sema`, but it results in fewer
bytes stored on the heap, without compromising accuracy and power of
compile error messages.
Miscellaneous:
* std.zig: string literals have more helpful result values for
reporting errors. There is now a lower level API and a higher level
API.
- side note: I noticed that the string literal logic needs some love.
There is some unnecessarily hacky code there.
* cut & pasted some TZIR logic that was in zir.zig to ir.zig. This
probably broke stuff and needs to get fixed.
* Removed type/Enum.zig, type/Union.zig, and type/Struct.zig. I don't
think this quite how this code will be organized. Need some more
careful planning about how to implement structs, unions, enums. They
need to be independent Decls, just like a top level function.
2021-03-15 23:38:38 -07:00
|
|
|
|
2022-07-15 16:05:06 +03:00
|
|
|
|
2021-04-28 22:43:26 -07:00
|
|
|
|
2021-05-02 15:06:32 -07:00
|
|
|
|
2022-07-15 16:05:06 +03:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-04-02 21:06:09 -07:00
|
|
|
|
2021-03-26 18:35:15 -07:00
|
|
|
|
2021-04-28 22:43:26 -07:00
|
|
|
|
2021-05-02 15:06:32 -07:00
|
|
|
|
2021-03-26 18:35:15 -07:00
|
|
|
|
|
|
|
|
|
2023-01-09 14:59:19 +00:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2022-07-28 17:28:37 +03:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-03-26 18:35:15 -07:00
|
|
|
|
|
|
|
|
|
2022-07-15 16:05:06 +03:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2022-07-14 21:26:01 +03:00
|
|
|
|
2023-06-22 18:46:56 +01:00
|
|
|
|
2022-07-15 11:38:16 +03:00
|
|
|
|
2021-03-26 18:35:15 -07:00
|
|
|
|
2022-07-28 17:28:37 +03:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-01-09 14:59:19 +00:00
|
|
|
|
2022-07-28 17:28:37 +03:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-01-09 14:59:19 +00:00
|
|
|
|
2022-07-28 17:28:37 +03:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-01-09 14:59:19 +00:00
|
|
|
|
2022-07-28 17:28:37 +03:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-01-09 14:59:19 +00:00
|
|
|
|
2022-07-28 17:28:37 +03:00
|
|
|
|
|
|
|
|
|
2021-09-14 21:58:22 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-06-25 02:26:48 +01:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-03-31 23:00:00 -07:00
|
|
|
|
2021-04-28 22:43:26 -07:00
|
|
|
|
2021-03-31 23:00:00 -07:00
|
|
|
|
2021-04-16 14:44:02 -07:00
|
|
|
|
2022-07-14 21:26:01 +03:00
|
|
|
|
2021-03-31 23:00:00 -07:00
|
|
|
|
2022-07-06 12:17:04 +03:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-04-28 22:43:26 -07:00
|
|
|
|
2021-04-16 14:44:02 -07:00
|
|
|
|
2023-01-09 14:59:19 +00:00
|
|
|
|
2022-07-15 11:38:16 +03:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2022-07-14 21:26:01 +03:00
|
|
|
|
2021-03-31 23:00:00 -07:00
|
|
|
|
|
|
|
|
|
2021-04-28 22:43:26 -07:00
|
|
|
|
2021-04-16 14:44:02 -07:00
|
|
|
|
2023-01-09 14:59:19 +00:00
|
|
|
|
|
|
|
|
|
2022-07-14 21:26:01 +03:00
|
|
|
|
2021-03-31 23:00:00 -07:00
|
|
|
|
|
|
|
|
|
2021-04-28 22:43:26 -07:00
|
|
|
|
2021-03-31 23:00:00 -07:00
|
|
|
|
|
|
|
|
|
2021-04-16 14:44:02 -07:00
|
|
|
|
Zir: eliminate `field_call_bind` and `field_call_bind_named`
This commit removes the `field_call_bind` and `field_call_bind_named` ZIR
instructions, replacing them with a `field_call` instruction which does the bind
and call in one.
`field_call_bind` is an unfortunate instruction. It's tied into one very
specific usage pattern - its result can only be used as a callee. This means
that it creates a value of a "pseudo-type" of sorts, `bound_fn` - this type used
to exist in Zig, but now we just hide it from the user and have AstGen ensure
it's only used in one way. This is quite silly - `Type` and `Value` should, as
much as possible, reflect real Zig types and values.
It makes sense to instead encode the `a.b()` syntax as its own ZIR instruction,
so that's what we do here. This commit introduces a new instruction,
`field_call`. It's like `call`, but rather than a callee ref, it contains a ref
to the object pointer (`&a` in `a.b()`) and the string field name (`b`). This
eliminates `bound_fn` from the language, and slightly decreases the size of
generated ZIR - stats below.
This commit does remove a few usages which used to be allowed:
- `@field(a, "b")()`
- `@call(.auto, a.b, .{})`
- `@call(.auto, @field(a, "b"), .{})`
These forms used to work just like `a.b()`, but are no longer allowed. I believe
this is the correct choice for a few reasons:
- `a.b()` is a purely *syntactic* form; for instance, `(a.b)()` is not valid.
This means it is *not* inconsistent to not allow it in these cases; the
special case here isn't "a field access as a callee", but rather this exact
syntactic form.
- The second argument to `@call` looks much more visually distinct from the
callee in standard call syntax. To me, this makes it seem strange for that
argument to not work like a normal expression in this context.
- A more practical argument: it's confusing! `@field` and `@call` are used in
very different contexts to standard function calls: the former normally hints
at some comptime machinery, and the latter that you want more precise control
over parts of a function call. In these contexts, you don't want implicit
arguments adding extra confusion: you want to be very explicit about what
you're doing.
Lastly, some stats. I mentioned before that this change slightly reduces the
size of ZIR - this is due to two instructions (`field_call_bind` then `call`)
being replaced with one (`field_call`). Here are some numbers:
+--------------+----------+----------+--------+
| File | Before | After | Change |
+--------------+----------+----------+--------+
| Sema.zig | 4.72M | 4.53M | -4% |
| AstGen.zig | 1.52M | 1.48M | -3% |
| hash_map.zig | 283.9K | 276.2K | -3% |
| math.zig | 312.6K | 305.3K | -2% |
+--------------+----------+----------+--------+
2023-05-13 17:10:05 +01:00
|
|
|
|
2021-04-02 21:06:09 -07:00
|
|
|
|
|
|
|
|
|
Zir: eliminate `field_call_bind` and `field_call_bind_named`
This commit removes the `field_call_bind` and `field_call_bind_named` ZIR
instructions, replacing them with a `field_call` instruction which does the bind
and call in one.
`field_call_bind` is an unfortunate instruction. It's tied into one very
specific usage pattern - its result can only be used as a callee. This means
that it creates a value of a "pseudo-type" of sorts, `bound_fn` - this type used
to exist in Zig, but now we just hide it from the user and have AstGen ensure
it's only used in one way. This is quite silly - `Type` and `Value` should, as
much as possible, reflect real Zig types and values.
It makes sense to instead encode the `a.b()` syntax as its own ZIR instruction,
so that's what we do here. This commit introduces a new instruction,
`field_call`. It's like `call`, but rather than a callee ref, it contains a ref
to the object pointer (`&a` in `a.b()`) and the string field name (`b`). This
eliminates `bound_fn` from the language, and slightly decreases the size of
generated ZIR - stats below.
This commit does remove a few usages which used to be allowed:
- `@field(a, "b")()`
- `@call(.auto, a.b, .{})`
- `@call(.auto, @field(a, "b"), .{})`
These forms used to work just like `a.b()`, but are no longer allowed. I believe
this is the correct choice for a few reasons:
- `a.b()` is a purely *syntactic* form; for instance, `(a.b)()` is not valid.
This means it is *not* inconsistent to not allow it in these cases; the
special case here isn't "a field access as a callee", but rather this exact
syntactic form.
- The second argument to `@call` looks much more visually distinct from the
callee in standard call syntax. To me, this makes it seem strange for that
argument to not work like a normal expression in this context.
- A more practical argument: it's confusing! `@field` and `@call` are used in
very different contexts to standard function calls: the former normally hints
at some comptime machinery, and the latter that you want more precise control
over parts of a function call. In these contexts, you don't want implicit
arguments adding extra confusion: you want to be very explicit about what
you're doing.
Lastly, some stats. I mentioned before that this change slightly reduces the
size of ZIR - this is due to two instructions (`field_call_bind` then `call`)
being replaced with one (`field_call`). Here are some numbers:
+--------------+----------+----------+--------+
| File | Before | After | Change |
+--------------+----------+----------+--------+
| Sema.zig | 4.72M | 4.53M | -4% |
| AstGen.zig | 1.52M | 1.48M | -3% |
| hash_map.zig | 283.9K | 276.2K | -3% |
| math.zig | 312.6K | 305.3K | -2% |
+--------------+----------+----------+--------+
2023-05-13 17:10:05 +01:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-04-02 21:06:09 -07:00
|
|
|
|
|
|
|
|
|
2022-07-14 21:26:01 +03:00
|
|
|
|
2023-06-22 18:46:56 +01:00
|
|
|
|
2022-07-15 11:38:16 +03:00
|
|
|
|
2021-03-31 23:00:00 -07:00
|
|
|
|
|
|
|
|
|
2021-04-28 22:43:26 -07:00
|
|
|
|
2021-04-16 14:44:02 -07:00
|
|
|
|
2022-07-14 21:26:01 +03:00
|
|
|
|
2021-03-31 23:00:00 -07:00
|
|
|
|
|
|
|
|
|
2021-04-28 22:43:26 -07:00
|
|
|
|
2021-04-16 14:44:02 -07:00
|
|
|
|
2023-01-09 14:59:19 +00:00
|
|
|
|
2022-07-14 21:26:01 +03:00
|
|
|
|
2021-03-31 23:00:00 -07:00
|
|
|
|
|
|
|
|
|
2021-04-28 22:43:26 -07:00
|
|
|
|
2021-04-16 14:44:02 -07:00
|
|
|
|
2023-01-09 14:59:19 +00:00
|
|
|
|
stage2 generics improvements: anytype and param type exprs
AstGen result locations now have a `coerced_ty` tag which is the same as
`ty` except it assumes that Sema will do a coercion, so it does not
redundantly add an `as` instruction into the ZIR code. This results in
cleaner ZIR and about a 14% reduction of ZIR bytes.
param and param_comptime ZIR instructions now have a block body for
their type expressions. This allows Sema to skip evaluation of the
block in the case that the parameter is comptime-provided. It also
allows a new mechanism to function: when evaluating type expressions of
generic functions, if it would depend on another parameter, it returns
`error.GenericPoison` which bubbles up and then is caught by the
param/param_comptime instruction and then handled.
This allows parameters to be evaluated independently so that the type
info for functions which have comptime or anytype parameters will still
have types populated for parameters that do not depend on values of
previous parameters (because evaluation of their param blocks will return
successfully instead of `error.GenericPoison`).
It also makes iteration over the block that contains function parameters
slightly more efficient since it now only contains the param
instructions.
Finally, it fixes the case where a generic function type expression contains
a function prototype. Formerly, this situation would cause shared state
to clobber each other; now it is in a proper tree structure so that
can't happen. This fix also required adding a field to Sema
`comptime_args_fn_inst` to make sure that the `comptime_args` field
passed into Sema is applied to the correct `func` instruction.
Source location for `node_offset_asm_ret_ty` is fixed; it was pointing at
the asm output name rather than the return type as intended.
Generic function instantiation is fixed, notably with respect to
parameter type expressions that depend on previous parameters, and with
respect to types which must be always comptime-known. This involves
passing all the comptime arguments at a callsite of a generic function,
and allowing the generic function semantic analysis to coerce the values
to the proper types (since it has access to the evaluated parameter type
expressions) and then decide based on the type whether the parameter is
runtime known or not. In the case of explicitly marked `comptime`
parameters, there is a check at the semantic analysis of the `call`
instruction.
Semantic analysis of `call` instructions does type coercion on the
arguments, which is needed both for generic functions and to make up for
using `coerced_ty` result locations (mentioned above).
Tasks left in this branch:
* Implement the memoization table.
* Add test coverage.
* Improve error reporting and source locations for compile errors.
2021-08-04 21:11:31 -07:00
|
|
|
|
|
|
|
|
|
2022-07-14 21:26:01 +03:00
|
|
|
|
2021-03-31 23:00:00 -07:00
|
|
|
|
2021-03-25 00:37:52 -07:00
|
|
|
|
2023-03-26 15:13:51 +03:00
|
|
|
|
2021-04-28 22:43:26 -07:00
|
|
|
|
2021-05-02 15:06:32 -07:00
|
|
|
|
2021-03-25 00:37:52 -07:00
|
|
|
|
2021-03-26 18:35:15 -07:00
|
|
|
|
2023-01-09 14:59:19 +00:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-03-26 15:13:51 +03:00
|
|
|
|
|
|
|
|
|
2023-01-09 14:59:19 +00:00
|
|
|
|
|
|
|
|
|
2023-03-26 15:13:51 +03:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-01-09 14:59:19 +00:00
|
|
|
|
2022-07-26 16:20:38 +03:00
|
|
|
|
|
|
|
|
|
2021-03-25 00:37:52 -07:00
|
|
|
|
|
|
|
|
|
2022-07-14 21:26:01 +03:00
|
|
|
|
2021-03-25 00:37:52 -07:00
|
|
|
|
2023-02-18 12:26:22 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-02-18 14:10:56 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-06-22 18:46:56 +01:00
|
|
|
|
2023-02-18 14:10:56 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
rework generic function calls
Abridged summary:
* Move `Module.Fn` into `InternPool`.
* Delete a lot of confusing and problematic `Sema` logic related to
generic function calls.
This commit removes `Module.Fn` and replaces it with two new
`InternPool.Tag` values:
* `func_decl` - corresponding to a function declared in the source
code. This one contains line/column numbers, zir_body_inst, etc.
* `func_instance` - one for each monomorphization of a generic
function. Contains a reference to the `func_decl` from whence the
instantiation came, along with the `comptime` parameter values (or
types in the case of `anytype`)
Since `InternPool` provides deduplication on these values, these fields
are now deleted from `Module`:
* `monomorphed_func_keys`
* `monomorphed_funcs`
* `align_stack_fns`
Instead of these, Sema logic for generic function instantiation now
unconditionally evaluates the function prototype expression for every
generic callsite. This is technically required in order for type
coercions to work. The previous code had some dubious, probably wrong
hacks to make things work, such as `hashUncoerced`. I'm not 100% sure
how we were able to eliminate that function and still pass all the
behavior tests, but I'm pretty sure things were still broken without
doing type coercion for every generic function call argument.
After the function prototype is evaluated, it produces a deduplicated
`func_instance` `InternPool.Index` which can then be used for the
generic function call.
Some other nice things made by this simplification are the removal of
`comptime_args_fn_inst` and `preallocated_new_func` from `Sema`, and the
messy logic associated with them.
I have not yet been able to measure the perf of this against master
branch. On one hand, it reduces memory usage and pointer chasing of the
most heavily used `InternPool` Tag - function bodies - but on the other
hand, it does evaluate function prototype expressions more than before.
We will soon find out.
2023-07-03 22:09:30 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-08-09 10:07:55 -04:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
rework generic function calls
Abridged summary:
* Move `Module.Fn` into `InternPool`.
* Delete a lot of confusing and problematic `Sema` logic related to
generic function calls.
This commit removes `Module.Fn` and replaces it with two new
`InternPool.Tag` values:
* `func_decl` - corresponding to a function declared in the source
code. This one contains line/column numbers, zir_body_inst, etc.
* `func_instance` - one for each monomorphization of a generic
function. Contains a reference to the `func_decl` from whence the
instantiation came, along with the `comptime` parameter values (or
types in the case of `anytype`)
Since `InternPool` provides deduplication on these values, these fields
are now deleted from `Module`:
* `monomorphed_func_keys`
* `monomorphed_funcs`
* `align_stack_fns`
Instead of these, Sema logic for generic function instantiation now
unconditionally evaluates the function prototype expression for every
generic callsite. This is technically required in order for type
coercions to work. The previous code had some dubious, probably wrong
hacks to make things work, such as `hashUncoerced`. I'm not 100% sure
how we were able to eliminate that function and still pass all the
behavior tests, but I'm pretty sure things were still broken without
doing type coercion for every generic function call argument.
After the function prototype is evaluated, it produces a deduplicated
`func_instance` `InternPool.Index` which can then be used for the
generic function call.
Some other nice things made by this simplification are the removal of
`comptime_args_fn_inst` and `preallocated_new_func` from `Sema`, and the
messy logic associated with them.
I have not yet been able to measure the perf of this against master
branch. On one hand, it reduces memory usage and pointer chasing of the
most heavily used `InternPool` Tag - function bodies - but on the other
hand, it does evaluate function prototype expressions more than before.
We will soon find out.
2023-07-03 22:09:30 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-07-18 17:50:41 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-07-18 18:34:14 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-07-18 17:50:41 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
rework generic function calls
Abridged summary:
* Move `Module.Fn` into `InternPool`.
* Delete a lot of confusing and problematic `Sema` logic related to
generic function calls.
This commit removes `Module.Fn` and replaces it with two new
`InternPool.Tag` values:
* `func_decl` - corresponding to a function declared in the source
code. This one contains line/column numbers, zir_body_inst, etc.
* `func_instance` - one for each monomorphization of a generic
function. Contains a reference to the `func_decl` from whence the
instantiation came, along with the `comptime` parameter values (or
types in the case of `anytype`)
Since `InternPool` provides deduplication on these values, these fields
are now deleted from `Module`:
* `monomorphed_func_keys`
* `monomorphed_funcs`
* `align_stack_fns`
Instead of these, Sema logic for generic function instantiation now
unconditionally evaluates the function prototype expression for every
generic callsite. This is technically required in order for type
coercions to work. The previous code had some dubious, probably wrong
hacks to make things work, such as `hashUncoerced`. I'm not 100% sure
how we were able to eliminate that function and still pass all the
behavior tests, but I'm pretty sure things were still broken without
doing type coercion for every generic function call argument.
After the function prototype is evaluated, it produces a deduplicated
`func_instance` `InternPool.Index` which can then be used for the
generic function call.
Some other nice things made by this simplification are the removal of
`comptime_args_fn_inst` and `preallocated_new_func` from `Sema`, and the
messy logic associated with them.
I have not yet been able to measure the perf of this against master
branch. On one hand, it reduces memory usage and pointer chasing of the
most heavily used `InternPool` Tag - function bodies - but on the other
hand, it does evaluate function prototype expressions more than before.
We will soon find out.
2023-07-03 22:09:30 -07:00
|
|
|
|
2021-03-26 23:46:37 -07:00
|
|
|
|
2021-04-28 22:43:26 -07:00
|
|
|
|
2021-05-02 15:06:32 -07:00
|
|
|
|
2021-03-26 23:46:37 -07:00
|
|
|
|
2022-07-14 21:26:01 +03:00
|
|
|
|
2021-03-26 23:46:37 -07:00
|
|
|
|
|
|
|
|
|
2021-04-28 22:43:26 -07:00
|
|
|
|
2021-05-02 15:06:32 -07:00
|
|
|
|
2021-03-26 23:46:37 -07:00
|
|
|
|
2022-07-14 21:26:01 +03:00
|
|
|
|
2021-03-26 23:46:37 -07:00
|
|
|
|
2021-03-31 23:00:00 -07:00
|
|
|
|
|
|
|
|
|
2021-04-28 22:43:26 -07:00
|
|
|
|
2021-05-02 15:06:32 -07:00
|
|
|
|
2021-03-31 23:00:00 -07:00
|
|
|
|
2022-07-14 21:26:01 +03:00
|
|
|
|
2021-03-31 23:00:00 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-04-28 22:43:26 -07:00
|
|
|
|
2021-05-02 15:06:32 -07:00
|
|
|
|
2021-03-31 23:00:00 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-08-30 19:22:04 -07:00
|
|
|
|
2021-03-31 23:00:00 -07:00
|
|
|
|
|
|
|
|
|
2023-01-09 14:59:19 +00:00
|
|
|
|
2021-03-31 23:00:00 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2022-07-14 21:26:01 +03:00
|
|
|
|
2021-03-31 23:00:00 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-04-28 22:43:26 -07:00
|
|
|
|
2021-05-02 15:06:32 -07:00
|
|
|
|
2021-03-31 23:00:00 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-08-30 19:22:04 -07:00
|
|
|
|
2021-03-31 23:00:00 -07:00
|
|
|
|
|
|
|
|
|
2023-01-09 14:59:19 +00:00
|
|
|
|
2021-03-31 23:00:00 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2022-07-14 21:26:01 +03:00
|
|
|
|
2021-03-31 23:00:00 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-05-28 04:31:56 +01:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2022-07-19 13:19:23 +03:00
|
|
|
|
|
|
|
|
|
2023-01-09 14:59:19 +00:00
|
|
|
|
2022-07-19 13:19:23 +03:00
|
|
|
|
2023-05-28 04:31:56 +01:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2022-07-19 13:19:23 +03:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-06-22 18:46:56 +01:00
|
|
|
|
2022-07-19 13:19:23 +03:00
|
|
|
|
|
|
|
|
|
2022-07-17 16:13:25 +03:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-01-09 14:59:19 +00:00
|
|
|
|
|
|
|
|
|
2022-07-17 16:13:25 +03:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-01-09 14:59:19 +00:00
|
|
|
|
|
|
|
|
|
2022-07-17 16:13:25 +03:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-01-09 14:59:19 +00:00
|
|
|
|
|
|
|
|
|
2022-07-17 16:13:25 +03:00
|
|
|
|
|
|
|
|
|
2021-03-31 23:00:00 -07:00
|
|
|
|
2021-04-28 22:43:26 -07:00
|
|
|
|
2021-04-16 14:44:02 -07:00
|
|
|
|
2023-01-09 14:59:19 +00:00
|
|
|
|
|
|
|
|
|
2022-07-14 21:26:01 +03:00
|
|
|
|
2021-03-31 23:00:00 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-04-28 22:43:26 -07:00
|
|
|
|
2021-04-16 14:44:02 -07:00
|
|
|
|
2023-01-09 14:59:19 +00:00
|
|
|
|
|
|
|
|
|
2022-07-14 21:26:01 +03:00
|
|
|
|
2021-03-31 23:00:00 -07:00
|
|
|
|
2022-07-19 20:20:57 +03:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-06-22 18:46:56 +01:00
|
|
|
|
2022-07-19 20:20:57 +03:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-04-23 18:28:46 -07:00
|
|
|
|
|
|
|
|
|
2021-04-28 22:43:26 -07:00
|
|
|
|
2021-04-23 18:28:46 -07:00
|
|
|
|
|
|
|
|
|
2022-07-14 21:26:01 +03:00
|
|
|
|
2021-04-23 18:28:46 -07:00
|
|
|
|
2021-05-04 14:40:59 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-01-09 14:59:19 +00:00
|
|
|
|
|
|
|
|
|
2021-05-04 14:40:59 -07:00
|
|
|
|
2022-07-14 21:26:01 +03:00
|
|
|
|
2023-06-22 18:46:56 +01:00
|
|
|
|
2022-07-15 11:38:16 +03:00
|
|
|
|
2021-05-04 14:40:59 -07:00
|
|
|
|
2021-10-12 21:26:59 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-01-09 14:59:19 +00:00
|
|
|
|
2022-07-14 21:26:01 +03:00
|
|
|
|
2021-10-12 21:26:59 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-01-09 14:59:19 +00:00
|
|
|
|
2022-07-14 21:26:01 +03:00
|
|
|
|
2021-10-12 21:26:59 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-01-09 14:59:19 +00:00
|
|
|
|
2022-07-14 21:26:01 +03:00
|
|
|
|
2021-10-12 21:26:59 -07:00
|
|
|
|
2022-07-06 11:36:39 +03:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2022-07-14 21:26:01 +03:00
|
|
|
|
2022-07-06 11:36:39 +03:00
|
|
|
|
2022-07-12 16:08:23 +03:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-01-09 14:59:19 +00:00
|
|
|
|
2022-07-12 16:08:23 +03:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-01-09 14:59:19 +00:00
|
|
|
|
2022-07-12 16:08:23 +03:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-01-09 14:59:19 +00:00
|
|
|
|
2022-07-12 16:08:23 +03:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-01-09 14:59:19 +00:00
|
|
|
|
2022-07-12 16:08:23 +03:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-01-09 14:59:19 +00:00
|
|
|
|
2022-07-12 16:08:23 +03:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-01-09 14:59:19 +00:00
|
|
|
|
2022-07-12 16:08:23 +03:00
|
|
|
|
|
|
|
|
|
2022-07-19 01:25:10 +03:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2022-07-22 17:15:15 +03:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2022-07-28 14:58:20 +03:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-01-09 14:59:19 +00:00
|
|
|
|
2022-07-28 14:58:20 +03:00
|
|
|
|
|
|
|
|
|
2022-10-18 13:45:06 +03:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
stage2: *WIP*: rework ZIR memory layout; overhaul source locations
The memory layout for ZIR instructions is completely reworked. See
zir.zig for those changes. Some new types:
* `zir.Code`: a "finished" set of ZIR instructions. Instead of allocating
each instruction independently, there is now a Tag and 8 bytes of
data available for all ZIR instructions. Small instructions fit
within these 8 bytes; larger ones use 4 bytes for an index into
`extra`. There is also `string_bytes` so that we can have 4 byte
references to strings. `zir.Inst.Tag` describes how to interpret
those 8 bytes of data.
- This is shared by all `Block` scopes.
* `Module.WipZirCode`: represents an in-progress `zir.Code`. In this
structure, the arrays are mutable, and get resized as we add/delete
things. There is extra state to keep track of things. This struct is
stored on the stack. Once it is finished, it produces an immutable
`zir.Code`, which will remain on the heap for the duration of a
function's existence.
- This is shared by all `GenZir` scopes.
* `Sema`: represents in-progress semantic analysis of a `zir.Code`.
This data is stored on the stack and is shared among all `Block`
scopes. It is now the main "self" argument to everything in the file
that was previously named `zir_sema.zig`.
Additionally, I moved some logic that was in `Module` into here.
`Module.Fn` now stores its parameter names inside the `zir.Code`,
instead of inside ZIR instructions. When the TZIR memory layout
reworking time comes, codegen will be able to reference this data
directly instead of duplicating it.
astgen.zig is (so far) almost entirely untouched, but nearly all of it
will need to be reworked to adhere to this new memory layout structure.
I have no benchmarks to report yet, as I am still working through
compile errors and fixing various things that I broke in this branch.
Overhaul of Source Locations:
Previously we used `usize` everywhere to mean byte offset, but sometimes
also mean other stuff. This was error prone and also made us do
unnecessary work, and store unnecessary bytes in memory.
Now there are more types involved into source locations, and more ways
to describe a source location.
* AllErrors.Message: embrace the assumption that files always have less
than 2 << 32 bytes.
* SrcLoc gets more complicated, to model more complicated source
locations.
* Introduce LazySrcLoc, which can model interesting source locations
with very little stored state. Useful for avoiding doing unnecessary
work when no compile errors occur.
Also, previously, we had `src: usize` on every ZIR instruction. This is
no longer the case. Each instruction now determines whether it even cares
about source location, and if so, how that source location is stored.
This requires more careful work inside `Sema`, but it results in fewer
bytes stored on the heap, without compromising accuracy and power of
compile error messages.
Miscellaneous:
* std.zig: string literals have more helpful result values for
reporting errors. There is now a lower level API and a higher level
API.
- side note: I noticed that the string literal logic needs some love.
There is some unnecessarily hacky code there.
* cut & pasted some TZIR logic that was in zir.zig to ir.zig. This
probably broke stuff and needs to get fixed.
* Removed type/Enum.zig, type/Union.zig, and type/Struct.zig. I don't
think this quite how this code will be organized. Need some more
careful planning about how to implement structs, unions, enums. They
need to be independent Decls, just like a top level function.
2021-03-15 23:38:38 -07:00
|
|
|
|
|
|
|
|
|
2021-09-14 21:58:22 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-10-29 00:37:25 +01:00
|
|
|
|
2021-09-14 21:58:22 -07:00
|
|
|
|
|
|
|
|
|
2022-07-14 21:26:01 +03:00
|
|
|
|
2021-09-14 21:58:22 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2022-07-14 21:26:01 +03:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2022-07-15 16:05:06 +03:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-07-18 17:50:41 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2022-07-15 16:05:06 +03:00
|
|
|
|
2021-09-14 21:58:22 -07:00
|
|
|
|
2022-07-15 11:38:16 +03:00
|
|
|
|
|
|
|
|
|
2022-07-14 21:26:01 +03:00
|
|
|
|
2022-07-15 11:38:16 +03:00
|
|
|
|
|
|
|
|
|
2022-07-15 16:05:06 +03:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2022-07-15 11:38:16 +03:00
|
|
|
|
2022-07-15 16:05:06 +03:00
|
|
|
|
|
|
|
|
|
2022-07-14 21:26:01 +03:00
|
|
|
|
2022-07-15 11:38:16 +03:00
|
|
|
|
2023-06-22 18:46:56 +01:00
|
|
|
|
2022-07-15 16:05:06 +03:00
|
|
|
|
2021-09-14 21:58:22 -07:00
|
|
|
|
stage2: *WIP*: rework ZIR memory layout; overhaul source locations
The memory layout for ZIR instructions is completely reworked. See
zir.zig for those changes. Some new types:
* `zir.Code`: a "finished" set of ZIR instructions. Instead of allocating
each instruction independently, there is now a Tag and 8 bytes of
data available for all ZIR instructions. Small instructions fit
within these 8 bytes; larger ones use 4 bytes for an index into
`extra`. There is also `string_bytes` so that we can have 4 byte
references to strings. `zir.Inst.Tag` describes how to interpret
those 8 bytes of data.
- This is shared by all `Block` scopes.
* `Module.WipZirCode`: represents an in-progress `zir.Code`. In this
structure, the arrays are mutable, and get resized as we add/delete
things. There is extra state to keep track of things. This struct is
stored on the stack. Once it is finished, it produces an immutable
`zir.Code`, which will remain on the heap for the duration of a
function's existence.
- This is shared by all `GenZir` scopes.
* `Sema`: represents in-progress semantic analysis of a `zir.Code`.
This data is stored on the stack and is shared among all `Block`
scopes. It is now the main "self" argument to everything in the file
that was previously named `zir_sema.zig`.
Additionally, I moved some logic that was in `Module` into here.
`Module.Fn` now stores its parameter names inside the `zir.Code`,
instead of inside ZIR instructions. When the TZIR memory layout
reworking time comes, codegen will be able to reference this data
directly instead of duplicating it.
astgen.zig is (so far) almost entirely untouched, but nearly all of it
will need to be reworked to adhere to this new memory layout structure.
I have no benchmarks to report yet, as I am still working through
compile errors and fixing various things that I broke in this branch.
Overhaul of Source Locations:
Previously we used `usize` everywhere to mean byte offset, but sometimes
also mean other stuff. This was error prone and also made us do
unnecessary work, and store unnecessary bytes in memory.
Now there are more types involved into source locations, and more ways
to describe a source location.
* AllErrors.Message: embrace the assumption that files always have less
than 2 << 32 bytes.
* SrcLoc gets more complicated, to model more complicated source
locations.
* Introduce LazySrcLoc, which can model interesting source locations
with very little stored state. Useful for avoiding doing unnecessary
work when no compile errors occur.
Also, previously, we had `src: usize` on every ZIR instruction. This is
no longer the case. Each instruction now determines whether it even cares
about source location, and if so, how that source location is stored.
This requires more careful work inside `Sema`, but it results in fewer
bytes stored on the heap, without compromising accuracy and power of
compile error messages.
Miscellaneous:
* std.zig: string literals have more helpful result values for
reporting errors. There is now a lower level API and a higher level
API.
- side note: I noticed that the string literal logic needs some love.
There is some unnecessarily hacky code there.
* cut & pasted some TZIR logic that was in zir.zig to ir.zig. This
probably broke stuff and needs to get fixed.
* Removed type/Enum.zig, type/Union.zig, and type/Struct.zig. I don't
think this quite how this code will be organized. Need some more
careful planning about how to implement structs, unions, enums. They
need to be independent Decls, just like a top level function.
2021-03-15 23:38:38 -07:00
|
|
|
|
|
|
|
|
|
2022-06-07 17:48:53 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2022-06-08 12:55:24 -07:00
|
|
|
|
2022-06-07 17:48:53 -07:00
|
|
|
|
2022-06-09 15:33:04 -07:00
|
|
|
|
2022-06-07 17:48:53 -07:00
|
|
|
|
|
|
|
|
|
stage2: *WIP*: rework ZIR memory layout; overhaul source locations
The memory layout for ZIR instructions is completely reworked. See
zir.zig for those changes. Some new types:
* `zir.Code`: a "finished" set of ZIR instructions. Instead of allocating
each instruction independently, there is now a Tag and 8 bytes of
data available for all ZIR instructions. Small instructions fit
within these 8 bytes; larger ones use 4 bytes for an index into
`extra`. There is also `string_bytes` so that we can have 4 byte
references to strings. `zir.Inst.Tag` describes how to interpret
those 8 bytes of data.
- This is shared by all `Block` scopes.
* `Module.WipZirCode`: represents an in-progress `zir.Code`. In this
structure, the arrays are mutable, and get resized as we add/delete
things. There is extra state to keep track of things. This struct is
stored on the stack. Once it is finished, it produces an immutable
`zir.Code`, which will remain on the heap for the duration of a
function's existence.
- This is shared by all `GenZir` scopes.
* `Sema`: represents in-progress semantic analysis of a `zir.Code`.
This data is stored on the stack and is shared among all `Block`
scopes. It is now the main "self" argument to everything in the file
that was previously named `zir_sema.zig`.
Additionally, I moved some logic that was in `Module` into here.
`Module.Fn` now stores its parameter names inside the `zir.Code`,
instead of inside ZIR instructions. When the TZIR memory layout
reworking time comes, codegen will be able to reference this data
directly instead of duplicating it.
astgen.zig is (so far) almost entirely untouched, but nearly all of it
will need to be reworked to adhere to this new memory layout structure.
I have no benchmarks to report yet, as I am still working through
compile errors and fixing various things that I broke in this branch.
Overhaul of Source Locations:
Previously we used `usize` everywhere to mean byte offset, but sometimes
also mean other stuff. This was error prone and also made us do
unnecessary work, and store unnecessary bytes in memory.
Now there are more types involved into source locations, and more ways
to describe a source location.
* AllErrors.Message: embrace the assumption that files always have less
than 2 << 32 bytes.
* SrcLoc gets more complicated, to model more complicated source
locations.
* Introduce LazySrcLoc, which can model interesting source locations
with very little stored state. Useful for avoiding doing unnecessary
work when no compile errors occur.
Also, previously, we had `src: usize` on every ZIR instruction. This is
no longer the case. Each instruction now determines whether it even cares
about source location, and if so, how that source location is stored.
This requires more careful work inside `Sema`, but it results in fewer
bytes stored on the heap, without compromising accuracy and power of
compile error messages.
Miscellaneous:
* std.zig: string literals have more helpful result values for
reporting errors. There is now a lower level API and a higher level
API.
- side note: I noticed that the string literal logic needs some love.
There is some unnecessarily hacky code there.
* cut & pasted some TZIR logic that was in zir.zig to ir.zig. This
probably broke stuff and needs to get fixed.
* Removed type/Enum.zig, type/Union.zig, and type/Struct.zig. I don't
think this quite how this code will be organized. Need some more
careful planning about how to implement structs, unions, enums. They
need to be independent Decls, just like a top level function.
2021-03-15 23:38:38 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-04-14 11:26:53 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
stage2: *WIP*: rework ZIR memory layout; overhaul source locations
The memory layout for ZIR instructions is completely reworked. See
zir.zig for those changes. Some new types:
* `zir.Code`: a "finished" set of ZIR instructions. Instead of allocating
each instruction independently, there is now a Tag and 8 bytes of
data available for all ZIR instructions. Small instructions fit
within these 8 bytes; larger ones use 4 bytes for an index into
`extra`. There is also `string_bytes` so that we can have 4 byte
references to strings. `zir.Inst.Tag` describes how to interpret
those 8 bytes of data.
- This is shared by all `Block` scopes.
* `Module.WipZirCode`: represents an in-progress `zir.Code`. In this
structure, the arrays are mutable, and get resized as we add/delete
things. There is extra state to keep track of things. This struct is
stored on the stack. Once it is finished, it produces an immutable
`zir.Code`, which will remain on the heap for the duration of a
function's existence.
- This is shared by all `GenZir` scopes.
* `Sema`: represents in-progress semantic analysis of a `zir.Code`.
This data is stored on the stack and is shared among all `Block`
scopes. It is now the main "self" argument to everything in the file
that was previously named `zir_sema.zig`.
Additionally, I moved some logic that was in `Module` into here.
`Module.Fn` now stores its parameter names inside the `zir.Code`,
instead of inside ZIR instructions. When the TZIR memory layout
reworking time comes, codegen will be able to reference this data
directly instead of duplicating it.
astgen.zig is (so far) almost entirely untouched, but nearly all of it
will need to be reworked to adhere to this new memory layout structure.
I have no benchmarks to report yet, as I am still working through
compile errors and fixing various things that I broke in this branch.
Overhaul of Source Locations:
Previously we used `usize` everywhere to mean byte offset, but sometimes
also mean other stuff. This was error prone and also made us do
unnecessary work, and store unnecessary bytes in memory.
Now there are more types involved into source locations, and more ways
to describe a source location.
* AllErrors.Message: embrace the assumption that files always have less
than 2 << 32 bytes.
* SrcLoc gets more complicated, to model more complicated source
locations.
* Introduce LazySrcLoc, which can model interesting source locations
with very little stored state. Useful for avoiding doing unnecessary
work when no compile errors occur.
Also, previously, we had `src: usize` on every ZIR instruction. This is
no longer the case. Each instruction now determines whether it even cares
about source location, and if so, how that source location is stored.
This requires more careful work inside `Sema`, but it results in fewer
bytes stored on the heap, without compromising accuracy and power of
compile error messages.
Miscellaneous:
* std.zig: string literals have more helpful result values for
reporting errors. There is now a lower level API and a higher level
API.
- side note: I noticed that the string literal logic needs some love.
There is some unnecessarily hacky code there.
* cut & pasted some TZIR logic that was in zir.zig to ir.zig. This
probably broke stuff and needs to get fixed.
* Removed type/Enum.zig, type/Union.zig, and type/Struct.zig. I don't
think this quite how this code will be organized. Need some more
careful planning about how to implement structs, unions, enums. They
need to be independent Decls, just like a top level function.
2021-03-15 23:38:38 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-03-17 22:54:56 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
stage2: *WIP*: rework ZIR memory layout; overhaul source locations
The memory layout for ZIR instructions is completely reworked. See
zir.zig for those changes. Some new types:
* `zir.Code`: a "finished" set of ZIR instructions. Instead of allocating
each instruction independently, there is now a Tag and 8 bytes of
data available for all ZIR instructions. Small instructions fit
within these 8 bytes; larger ones use 4 bytes for an index into
`extra`. There is also `string_bytes` so that we can have 4 byte
references to strings. `zir.Inst.Tag` describes how to interpret
those 8 bytes of data.
- This is shared by all `Block` scopes.
* `Module.WipZirCode`: represents an in-progress `zir.Code`. In this
structure, the arrays are mutable, and get resized as we add/delete
things. There is extra state to keep track of things. This struct is
stored on the stack. Once it is finished, it produces an immutable
`zir.Code`, which will remain on the heap for the duration of a
function's existence.
- This is shared by all `GenZir` scopes.
* `Sema`: represents in-progress semantic analysis of a `zir.Code`.
This data is stored on the stack and is shared among all `Block`
scopes. It is now the main "self" argument to everything in the file
that was previously named `zir_sema.zig`.
Additionally, I moved some logic that was in `Module` into here.
`Module.Fn` now stores its parameter names inside the `zir.Code`,
instead of inside ZIR instructions. When the TZIR memory layout
reworking time comes, codegen will be able to reference this data
directly instead of duplicating it.
astgen.zig is (so far) almost entirely untouched, but nearly all of it
will need to be reworked to adhere to this new memory layout structure.
I have no benchmarks to report yet, as I am still working through
compile errors and fixing various things that I broke in this branch.
Overhaul of Source Locations:
Previously we used `usize` everywhere to mean byte offset, but sometimes
also mean other stuff. This was error prone and also made us do
unnecessary work, and store unnecessary bytes in memory.
Now there are more types involved into source locations, and more ways
to describe a source location.
* AllErrors.Message: embrace the assumption that files always have less
than 2 << 32 bytes.
* SrcLoc gets more complicated, to model more complicated source
locations.
* Introduce LazySrcLoc, which can model interesting source locations
with very little stored state. Useful for avoiding doing unnecessary
work when no compile errors occur.
Also, previously, we had `src: usize` on every ZIR instruction. This is
no longer the case. Each instruction now determines whether it even cares
about source location, and if so, how that source location is stored.
This requires more careful work inside `Sema`, but it results in fewer
bytes stored on the heap, without compromising accuracy and power of
compile error messages.
Miscellaneous:
* std.zig: string literals have more helpful result values for
reporting errors. There is now a lower level API and a higher level
API.
- side note: I noticed that the string literal logic needs some love.
There is some unnecessarily hacky code there.
* cut & pasted some TZIR logic that was in zir.zig to ir.zig. This
probably broke stuff and needs to get fixed.
* Removed type/Enum.zig, type/Union.zig, and type/Struct.zig. I don't
think this quite how this code will be organized. Need some more
careful planning about how to implement structs, unions, enums. They
need to be independent Decls, just like a top level function.
2021-03-15 23:38:38 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2022-06-07 17:48:53 -07:00
|
|
|
|
2022-12-11 20:41:42 +01:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2022-07-15 16:05:06 +03:00
|
|
|
|
2021-04-02 21:06:09 -07:00
|
|
|
|
2022-07-15 16:05:06 +03:00
|
|
|
|
stage2: *WIP*: rework ZIR memory layout; overhaul source locations
The memory layout for ZIR instructions is completely reworked. See
zir.zig for those changes. Some new types:
* `zir.Code`: a "finished" set of ZIR instructions. Instead of allocating
each instruction independently, there is now a Tag and 8 bytes of
data available for all ZIR instructions. Small instructions fit
within these 8 bytes; larger ones use 4 bytes for an index into
`extra`. There is also `string_bytes` so that we can have 4 byte
references to strings. `zir.Inst.Tag` describes how to interpret
those 8 bytes of data.
- This is shared by all `Block` scopes.
* `Module.WipZirCode`: represents an in-progress `zir.Code`. In this
structure, the arrays are mutable, and get resized as we add/delete
things. There is extra state to keep track of things. This struct is
stored on the stack. Once it is finished, it produces an immutable
`zir.Code`, which will remain on the heap for the duration of a
function's existence.
- This is shared by all `GenZir` scopes.
* `Sema`: represents in-progress semantic analysis of a `zir.Code`.
This data is stored on the stack and is shared among all `Block`
scopes. It is now the main "self" argument to everything in the file
that was previously named `zir_sema.zig`.
Additionally, I moved some logic that was in `Module` into here.
`Module.Fn` now stores its parameter names inside the `zir.Code`,
instead of inside ZIR instructions. When the TZIR memory layout
reworking time comes, codegen will be able to reference this data
directly instead of duplicating it.
astgen.zig is (so far) almost entirely untouched, but nearly all of it
will need to be reworked to adhere to this new memory layout structure.
I have no benchmarks to report yet, as I am still working through
compile errors and fixing various things that I broke in this branch.
Overhaul of Source Locations:
Previously we used `usize` everywhere to mean byte offset, but sometimes
also mean other stuff. This was error prone and also made us do
unnecessary work, and store unnecessary bytes in memory.
Now there are more types involved into source locations, and more ways
to describe a source location.
* AllErrors.Message: embrace the assumption that files always have less
than 2 << 32 bytes.
* SrcLoc gets more complicated, to model more complicated source
locations.
* Introduce LazySrcLoc, which can model interesting source locations
with very little stored state. Useful for avoiding doing unnecessary
work when no compile errors occur.
Also, previously, we had `src: usize` on every ZIR instruction. This is
no longer the case. Each instruction now determines whether it even cares
about source location, and if so, how that source location is stored.
This requires more careful work inside `Sema`, but it results in fewer
bytes stored on the heap, without compromising accuracy and power of
compile error messages.
Miscellaneous:
* std.zig: string literals have more helpful result values for
reporting errors. There is now a lower level API and a higher level
API.
- side note: I noticed that the string literal logic needs some love.
There is some unnecessarily hacky code there.
* cut & pasted some TZIR logic that was in zir.zig to ir.zig. This
probably broke stuff and needs to get fixed.
* Removed type/Enum.zig, type/Union.zig, and type/Struct.zig. I don't
think this quite how this code will be organized. Need some more
careful planning about how to implement structs, unions, enums. They
need to be independent Decls, just like a top level function.
2021-03-15 23:38:38 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-03-20 17:09:06 -07:00
|
|
|
|
2022-07-28 17:28:37 +03:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
stage2: *WIP*: rework ZIR memory layout; overhaul source locations
The memory layout for ZIR instructions is completely reworked. See
zir.zig for those changes. Some new types:
* `zir.Code`: a "finished" set of ZIR instructions. Instead of allocating
each instruction independently, there is now a Tag and 8 bytes of
data available for all ZIR instructions. Small instructions fit
within these 8 bytes; larger ones use 4 bytes for an index into
`extra`. There is also `string_bytes` so that we can have 4 byte
references to strings. `zir.Inst.Tag` describes how to interpret
those 8 bytes of data.
- This is shared by all `Block` scopes.
* `Module.WipZirCode`: represents an in-progress `zir.Code`. In this
structure, the arrays are mutable, and get resized as we add/delete
things. There is extra state to keep track of things. This struct is
stored on the stack. Once it is finished, it produces an immutable
`zir.Code`, which will remain on the heap for the duration of a
function's existence.
- This is shared by all `GenZir` scopes.
* `Sema`: represents in-progress semantic analysis of a `zir.Code`.
This data is stored on the stack and is shared among all `Block`
scopes. It is now the main "self" argument to everything in the file
that was previously named `zir_sema.zig`.
Additionally, I moved some logic that was in `Module` into here.
`Module.Fn` now stores its parameter names inside the `zir.Code`,
instead of inside ZIR instructions. When the TZIR memory layout
reworking time comes, codegen will be able to reference this data
directly instead of duplicating it.
astgen.zig is (so far) almost entirely untouched, but nearly all of it
will need to be reworked to adhere to this new memory layout structure.
I have no benchmarks to report yet, as I am still working through
compile errors and fixing various things that I broke in this branch.
Overhaul of Source Locations:
Previously we used `usize` everywhere to mean byte offset, but sometimes
also mean other stuff. This was error prone and also made us do
unnecessary work, and store unnecessary bytes in memory.
Now there are more types involved into source locations, and more ways
to describe a source location.
* AllErrors.Message: embrace the assumption that files always have less
than 2 << 32 bytes.
* SrcLoc gets more complicated, to model more complicated source
locations.
* Introduce LazySrcLoc, which can model interesting source locations
with very little stored state. Useful for avoiding doing unnecessary
work when no compile errors occur.
Also, previously, we had `src: usize` on every ZIR instruction. This is
no longer the case. Each instruction now determines whether it even cares
about source location, and if so, how that source location is stored.
This requires more careful work inside `Sema`, but it results in fewer
bytes stored on the heap, without compromising accuracy and power of
compile error messages.
Miscellaneous:
* std.zig: string literals have more helpful result values for
reporting errors. There is now a lower level API and a higher level
API.
- side note: I noticed that the string literal logic needs some love.
There is some unnecessarily hacky code there.
* cut & pasted some TZIR logic that was in zir.zig to ir.zig. This
probably broke stuff and needs to get fixed.
* Removed type/Enum.zig, type/Union.zig, and type/Struct.zig. I don't
think this quite how this code will be organized. Need some more
careful planning about how to implement structs, unions, enums. They
need to be independent Decls, just like a top level function.
2021-03-15 23:38:38 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-03-20 17:09:06 -07:00
|
|
|
|
stage2: *WIP*: rework ZIR memory layout; overhaul source locations
The memory layout for ZIR instructions is completely reworked. See
zir.zig for those changes. Some new types:
* `zir.Code`: a "finished" set of ZIR instructions. Instead of allocating
each instruction independently, there is now a Tag and 8 bytes of
data available for all ZIR instructions. Small instructions fit
within these 8 bytes; larger ones use 4 bytes for an index into
`extra`. There is also `string_bytes` so that we can have 4 byte
references to strings. `zir.Inst.Tag` describes how to interpret
those 8 bytes of data.
- This is shared by all `Block` scopes.
* `Module.WipZirCode`: represents an in-progress `zir.Code`. In this
structure, the arrays are mutable, and get resized as we add/delete
things. There is extra state to keep track of things. This struct is
stored on the stack. Once it is finished, it produces an immutable
`zir.Code`, which will remain on the heap for the duration of a
function's existence.
- This is shared by all `GenZir` scopes.
* `Sema`: represents in-progress semantic analysis of a `zir.Code`.
This data is stored on the stack and is shared among all `Block`
scopes. It is now the main "self" argument to everything in the file
that was previously named `zir_sema.zig`.
Additionally, I moved some logic that was in `Module` into here.
`Module.Fn` now stores its parameter names inside the `zir.Code`,
instead of inside ZIR instructions. When the TZIR memory layout
reworking time comes, codegen will be able to reference this data
directly instead of duplicating it.
astgen.zig is (so far) almost entirely untouched, but nearly all of it
will need to be reworked to adhere to this new memory layout structure.
I have no benchmarks to report yet, as I am still working through
compile errors and fixing various things that I broke in this branch.
Overhaul of Source Locations:
Previously we used `usize` everywhere to mean byte offset, but sometimes
also mean other stuff. This was error prone and also made us do
unnecessary work, and store unnecessary bytes in memory.
Now there are more types involved into source locations, and more ways
to describe a source location.
* AllErrors.Message: embrace the assumption that files always have less
than 2 << 32 bytes.
* SrcLoc gets more complicated, to model more complicated source
locations.
* Introduce LazySrcLoc, which can model interesting source locations
with very little stored state. Useful for avoiding doing unnecessary
work when no compile errors occur.
Also, previously, we had `src: usize` on every ZIR instruction. This is
no longer the case. Each instruction now determines whether it even cares
about source location, and if so, how that source location is stored.
This requires more careful work inside `Sema`, but it results in fewer
bytes stored on the heap, without compromising accuracy and power of
compile error messages.
Miscellaneous:
* std.zig: string literals have more helpful result values for
reporting errors. There is now a lower level API and a higher level
API.
- side note: I noticed that the string literal logic needs some love.
There is some unnecessarily hacky code there.
* cut & pasted some TZIR logic that was in zir.zig to ir.zig. This
probably broke stuff and needs to get fixed.
* Removed type/Enum.zig, type/Union.zig, and type/Struct.zig. I don't
think this quite how this code will be organized. Need some more
careful planning about how to implement structs, unions, enums. They
need to be independent Decls, just like a top level function.
2021-03-15 23:38:38 -07:00
|
|
|
|
2021-03-20 17:09:06 -07:00
|
|
|
|
2021-09-14 21:58:22 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-06-25 02:26:48 +01:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
stage2: *WIP*: rework ZIR memory layout; overhaul source locations
The memory layout for ZIR instructions is completely reworked. See
zir.zig for those changes. Some new types:
* `zir.Code`: a "finished" set of ZIR instructions. Instead of allocating
each instruction independently, there is now a Tag and 8 bytes of
data available for all ZIR instructions. Small instructions fit
within these 8 bytes; larger ones use 4 bytes for an index into
`extra`. There is also `string_bytes` so that we can have 4 byte
references to strings. `zir.Inst.Tag` describes how to interpret
those 8 bytes of data.
- This is shared by all `Block` scopes.
* `Module.WipZirCode`: represents an in-progress `zir.Code`. In this
structure, the arrays are mutable, and get resized as we add/delete
things. There is extra state to keep track of things. This struct is
stored on the stack. Once it is finished, it produces an immutable
`zir.Code`, which will remain on the heap for the duration of a
function's existence.
- This is shared by all `GenZir` scopes.
* `Sema`: represents in-progress semantic analysis of a `zir.Code`.
This data is stored on the stack and is shared among all `Block`
scopes. It is now the main "self" argument to everything in the file
that was previously named `zir_sema.zig`.
Additionally, I moved some logic that was in `Module` into here.
`Module.Fn` now stores its parameter names inside the `zir.Code`,
instead of inside ZIR instructions. When the TZIR memory layout
reworking time comes, codegen will be able to reference this data
directly instead of duplicating it.
astgen.zig is (so far) almost entirely untouched, but nearly all of it
will need to be reworked to adhere to this new memory layout structure.
I have no benchmarks to report yet, as I am still working through
compile errors and fixing various things that I broke in this branch.
Overhaul of Source Locations:
Previously we used `usize` everywhere to mean byte offset, but sometimes
also mean other stuff. This was error prone and also made us do
unnecessary work, and store unnecessary bytes in memory.
Now there are more types involved into source locations, and more ways
to describe a source location.
* AllErrors.Message: embrace the assumption that files always have less
than 2 << 32 bytes.
* SrcLoc gets more complicated, to model more complicated source
locations.
* Introduce LazySrcLoc, which can model interesting source locations
with very little stored state. Useful for avoiding doing unnecessary
work when no compile errors occur.
Also, previously, we had `src: usize` on every ZIR instruction. This is
no longer the case. Each instruction now determines whether it even cares
about source location, and if so, how that source location is stored.
This requires more careful work inside `Sema`, but it results in fewer
bytes stored on the heap, without compromising accuracy and power of
compile error messages.
Miscellaneous:
* std.zig: string literals have more helpful result values for
reporting errors. There is now a lower level API and a higher level
API.
- side note: I noticed that the string literal logic needs some love.
There is some unnecessarily hacky code there.
* cut & pasted some TZIR logic that was in zir.zig to ir.zig. This
probably broke stuff and needs to get fixed.
* Removed type/Enum.zig, type/Union.zig, and type/Struct.zig. I don't
think this quite how this code will be organized. Need some more
careful planning about how to implement structs, unions, enums. They
need to be independent Decls, just like a top level function.
2021-03-15 23:38:38 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-03-20 17:09:06 -07:00
|
|
|
|
2022-07-06 12:17:04 +03:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
stage2: *WIP*: rework ZIR memory layout; overhaul source locations
The memory layout for ZIR instructions is completely reworked. See
zir.zig for those changes. Some new types:
* `zir.Code`: a "finished" set of ZIR instructions. Instead of allocating
each instruction independently, there is now a Tag and 8 bytes of
data available for all ZIR instructions. Small instructions fit
within these 8 bytes; larger ones use 4 bytes for an index into
`extra`. There is also `string_bytes` so that we can have 4 byte
references to strings. `zir.Inst.Tag` describes how to interpret
those 8 bytes of data.
- This is shared by all `Block` scopes.
* `Module.WipZirCode`: represents an in-progress `zir.Code`. In this
structure, the arrays are mutable, and get resized as we add/delete
things. There is extra state to keep track of things. This struct is
stored on the stack. Once it is finished, it produces an immutable
`zir.Code`, which will remain on the heap for the duration of a
function's existence.
- This is shared by all `GenZir` scopes.
* `Sema`: represents in-progress semantic analysis of a `zir.Code`.
This data is stored on the stack and is shared among all `Block`
scopes. It is now the main "self" argument to everything in the file
that was previously named `zir_sema.zig`.
Additionally, I moved some logic that was in `Module` into here.
`Module.Fn` now stores its parameter names inside the `zir.Code`,
instead of inside ZIR instructions. When the TZIR memory layout
reworking time comes, codegen will be able to reference this data
directly instead of duplicating it.
astgen.zig is (so far) almost entirely untouched, but nearly all of it
will need to be reworked to adhere to this new memory layout structure.
I have no benchmarks to report yet, as I am still working through
compile errors and fixing various things that I broke in this branch.
Overhaul of Source Locations:
Previously we used `usize` everywhere to mean byte offset, but sometimes
also mean other stuff. This was error prone and also made us do
unnecessary work, and store unnecessary bytes in memory.
Now there are more types involved into source locations, and more ways
to describe a source location.
* AllErrors.Message: embrace the assumption that files always have less
than 2 << 32 bytes.
* SrcLoc gets more complicated, to model more complicated source
locations.
* Introduce LazySrcLoc, which can model interesting source locations
with very little stored state. Useful for avoiding doing unnecessary
work when no compile errors occur.
Also, previously, we had `src: usize` on every ZIR instruction. This is
no longer the case. Each instruction now determines whether it even cares
about source location, and if so, how that source location is stored.
This requires more careful work inside `Sema`, but it results in fewer
bytes stored on the heap, without compromising accuracy and power of
compile error messages.
Miscellaneous:
* std.zig: string literals have more helpful result values for
reporting errors. There is now a lower level API and a higher level
API.
- side note: I noticed that the string literal logic needs some love.
There is some unnecessarily hacky code there.
* cut & pasted some TZIR logic that was in zir.zig to ir.zig. This
probably broke stuff and needs to get fixed.
* Removed type/Enum.zig, type/Union.zig, and type/Struct.zig. I don't
think this quite how this code will be organized. Need some more
careful planning about how to implement structs, unions, enums. They
need to be independent Decls, just like a top level function.
2021-03-15 23:38:38 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-03-20 17:09:06 -07:00
|
|
|
|
2021-03-17 22:54:56 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-03-20 17:09:06 -07:00
|
|
|
|
2021-04-02 21:06:09 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
Zir: eliminate `field_call_bind` and `field_call_bind_named`
This commit removes the `field_call_bind` and `field_call_bind_named` ZIR
instructions, replacing them with a `field_call` instruction which does the bind
and call in one.
`field_call_bind` is an unfortunate instruction. It's tied into one very
specific usage pattern - its result can only be used as a callee. This means
that it creates a value of a "pseudo-type" of sorts, `bound_fn` - this type used
to exist in Zig, but now we just hide it from the user and have AstGen ensure
it's only used in one way. This is quite silly - `Type` and `Value` should, as
much as possible, reflect real Zig types and values.
It makes sense to instead encode the `a.b()` syntax as its own ZIR instruction,
so that's what we do here. This commit introduces a new instruction,
`field_call`. It's like `call`, but rather than a callee ref, it contains a ref
to the object pointer (`&a` in `a.b()`) and the string field name (`b`). This
eliminates `bound_fn` from the language, and slightly decreases the size of
generated ZIR - stats below.
This commit does remove a few usages which used to be allowed:
- `@field(a, "b")()`
- `@call(.auto, a.b, .{})`
- `@call(.auto, @field(a, "b"), .{})`
These forms used to work just like `a.b()`, but are no longer allowed. I believe
this is the correct choice for a few reasons:
- `a.b()` is a purely *syntactic* form; for instance, `(a.b)()` is not valid.
This means it is *not* inconsistent to not allow it in these cases; the
special case here isn't "a field access as a callee", but rather this exact
syntactic form.
- The second argument to `@call` looks much more visually distinct from the
callee in standard call syntax. To me, this makes it seem strange for that
argument to not work like a normal expression in this context.
- A more practical argument: it's confusing! `@field` and `@call` are used in
very different contexts to standard function calls: the former normally hints
at some comptime machinery, and the latter that you want more precise control
over parts of a function call. In these contexts, you don't want implicit
arguments adding extra confusion: you want to be very explicit about what
you're doing.
Lastly, some stats. I mentioned before that this change slightly reduces the
size of ZIR - this is due to two instructions (`field_call_bind` then `call`)
being replaced with one (`field_call`). Here are some numbers:
+--------------+----------+----------+--------+
| File | Before | After | Change |
+--------------+----------+----------+--------+
| Sema.zig | 4.72M | 4.53M | -4% |
| AstGen.zig | 1.52M | 1.48M | -3% |
| hash_map.zig | 283.9K | 276.2K | -3% |
| math.zig | 312.6K | 305.3K | -2% |
+--------------+----------+----------+--------+
2023-05-13 17:10:05 +01:00
|
|
|
|
|
|
|
|
|
2021-03-17 22:54:56 -07:00
|
|
|
|
2021-03-20 17:09:06 -07:00
|
|
|
|
2021-03-17 22:54:56 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-03-20 17:09:06 -07:00
|
|
|
|
2021-03-17 22:54:56 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-03-20 17:09:06 -07:00
|
|
|
|
2021-03-17 22:54:56 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-03-20 17:09:06 -07:00
|
|
|
|
2021-03-17 22:54:56 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-03-20 17:09:06 -07:00
|
|
|
|
2021-03-21 19:23:12 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-09-24 13:39:20 -04:00
|
|
|
|
2021-03-21 19:23:12 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-09-24 13:39:20 -04:00
|
|
|
|
2021-03-21 19:23:12 -07:00
|
|
|
|
|
|
|
|
|
stage2: guidance on how to implement switch expressions
Here's what I think the ZIR should be. AstGen is not yet implemented to
match this, and the main implementation of analyzeSwitch in Sema is not
yet implemented to match it either.
Here are some example byte size reductions from master branch, with the
ZIR memory layout from this commit:
```
switch (foo) {
a => 1,
b => 2,
c => 3,
d => 4,
}
```
184 bytes (master) => 40 bytes (this branch)
```
switch (foo) {
a, b => 1,
c..d, e, f => 2,
g => 3,
else => 4,
}
```
240 bytes (master) => 80 bytes (this branch)
2021-03-28 23:12:26 -07:00
|
|
|
|
|
|
|
|
|
2021-09-24 13:39:20 -04:00
|
|
|
|
stage2: guidance on how to implement switch expressions
Here's what I think the ZIR should be. AstGen is not yet implemented to
match this, and the main implementation of analyzeSwitch in Sema is not
yet implemented to match it either.
Here are some example byte size reductions from master branch, with the
ZIR memory layout from this commit:
```
switch (foo) {
a => 1,
b => 2,
c => 3,
d => 4,
}
```
184 bytes (master) => 40 bytes (this branch)
```
switch (foo) {
a, b => 1,
c..d, e, f => 2,
g => 3,
else => 4,
}
```
240 bytes (master) => 80 bytes (this branch)
2021-03-28 23:12:26 -07:00
|
|
|
|
|
|
|
|
|
2021-03-29 21:59:08 -07:00
|
|
|
|
|
|
|
|
|
2021-09-24 13:39:20 -04:00
|
|
|
|
2021-03-29 21:59:08 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-09-24 13:39:20 -04:00
|
|
|
|
2021-03-29 21:59:08 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2022-07-19 13:19:23 +03:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-05-28 04:31:56 +01:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2022-07-17 16:13:25 +03:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-03-31 21:36:32 -07:00
|
|
|
|
|
|
|
|
|
2021-09-24 13:39:20 -04:00
|
|
|
|
2021-03-31 21:36:32 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-09-24 13:39:20 -04:00
|
|
|
|
2021-03-31 21:36:32 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2022-07-19 20:20:57 +03:00
|
|
|
|
|
|
|
|
|
2021-04-23 18:28:46 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-05-04 14:40:59 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-10-12 21:26:59 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2022-07-06 11:36:39 +03:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2022-07-12 16:08:23 +03:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2022-07-19 01:25:10 +03:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2022-07-22 17:15:15 +03:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2022-07-28 14:58:20 +03:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2022-10-18 13:45:06 +03:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-02-18 12:26:22 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-02-18 14:10:56 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
rework generic function calls
Abridged summary:
* Move `Module.Fn` into `InternPool`.
* Delete a lot of confusing and problematic `Sema` logic related to
generic function calls.
This commit removes `Module.Fn` and replaces it with two new
`InternPool.Tag` values:
* `func_decl` - corresponding to a function declared in the source
code. This one contains line/column numbers, zir_body_inst, etc.
* `func_instance` - one for each monomorphization of a generic
function. Contains a reference to the `func_decl` from whence the
instantiation came, along with the `comptime` parameter values (or
types in the case of `anytype`)
Since `InternPool` provides deduplication on these values, these fields
are now deleted from `Module`:
* `monomorphed_func_keys`
* `monomorphed_funcs`
* `align_stack_fns`
Instead of these, Sema logic for generic function instantiation now
unconditionally evaluates the function prototype expression for every
generic callsite. This is technically required in order for type
coercions to work. The previous code had some dubious, probably wrong
hacks to make things work, such as `hashUncoerced`. I'm not 100% sure
how we were able to eliminate that function and still pass all the
behavior tests, but I'm pretty sure things were still broken without
doing type coercion for every generic function call argument.
After the function prototype is evaluated, it produces a deduplicated
`func_instance` `InternPool.Index` which can then be used for the
generic function call.
Some other nice things made by this simplification are the removal of
`comptime_args_fn_inst` and `preallocated_new_func` from `Sema`, and the
messy logic associated with them.
I have not yet been able to measure the perf of this against master
branch. On one hand, it reduces memory usage and pointer chasing of the
most heavily used `InternPool` Tag - function bodies - but on the other
hand, it does evaluate function prototype expressions more than before.
We will soon find out.
2023-07-03 22:09:30 -07:00
|
|
|
|
|
|
|
|
|
2023-07-18 17:12:19 -07:00
|
|
|
|
rework generic function calls
Abridged summary:
* Move `Module.Fn` into `InternPool`.
* Delete a lot of confusing and problematic `Sema` logic related to
generic function calls.
This commit removes `Module.Fn` and replaces it with two new
`InternPool.Tag` values:
* `func_decl` - corresponding to a function declared in the source
code. This one contains line/column numbers, zir_body_inst, etc.
* `func_instance` - one for each monomorphization of a generic
function. Contains a reference to the `func_decl` from whence the
instantiation came, along with the `comptime` parameter values (or
types in the case of `anytype`)
Since `InternPool` provides deduplication on these values, these fields
are now deleted from `Module`:
* `monomorphed_func_keys`
* `monomorphed_funcs`
* `align_stack_fns`
Instead of these, Sema logic for generic function instantiation now
unconditionally evaluates the function prototype expression for every
generic callsite. This is technically required in order for type
coercions to work. The previous code had some dubious, probably wrong
hacks to make things work, such as `hashUncoerced`. I'm not 100% sure
how we were able to eliminate that function and still pass all the
behavior tests, but I'm pretty sure things were still broken without
doing type coercion for every generic function call argument.
After the function prototype is evaluated, it produces a deduplicated
`func_instance` `InternPool.Index` which can then be used for the
generic function call.
Some other nice things made by this simplification are the removal of
`comptime_args_fn_inst` and `preallocated_new_func` from `Sema`, and the
messy logic associated with them.
I have not yet been able to measure the perf of this against master
branch. On one hand, it reduces memory usage and pointer chasing of the
most heavily used `InternPool` Tag - function bodies - but on the other
hand, it does evaluate function prototype expressions more than before.
We will soon find out.
2023-07-03 22:09:30 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-07-18 17:12:19 -07:00
|
|
|
|
rework generic function calls
Abridged summary:
* Move `Module.Fn` into `InternPool`.
* Delete a lot of confusing and problematic `Sema` logic related to
generic function calls.
This commit removes `Module.Fn` and replaces it with two new
`InternPool.Tag` values:
* `func_decl` - corresponding to a function declared in the source
code. This one contains line/column numbers, zir_body_inst, etc.
* `func_instance` - one for each monomorphization of a generic
function. Contains a reference to the `func_decl` from whence the
instantiation came, along with the `comptime` parameter values (or
types in the case of `anytype`)
Since `InternPool` provides deduplication on these values, these fields
are now deleted from `Module`:
* `monomorphed_func_keys`
* `monomorphed_funcs`
* `align_stack_fns`
Instead of these, Sema logic for generic function instantiation now
unconditionally evaluates the function prototype expression for every
generic callsite. This is technically required in order for type
coercions to work. The previous code had some dubious, probably wrong
hacks to make things work, such as `hashUncoerced`. I'm not 100% sure
how we were able to eliminate that function and still pass all the
behavior tests, but I'm pretty sure things were still broken without
doing type coercion for every generic function call argument.
After the function prototype is evaluated, it produces a deduplicated
`func_instance` `InternPool.Index` which can then be used for the
generic function call.
Some other nice things made by this simplification are the removal of
`comptime_args_fn_inst` and `preallocated_new_func` from `Sema`, and the
messy logic associated with them.
I have not yet been able to measure the perf of this against master
branch. On one hand, it reduces memory usage and pointer chasing of the
most heavily used `InternPool` Tag - function bodies - but on the other
hand, it does evaluate function prototype expressions more than before.
We will soon find out.
2023-07-03 22:09:30 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-03-17 00:56:08 -07:00
|
|
|
|
2022-06-07 17:48:53 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-03-17 22:54:56 -07:00
|
|
|
|
2023-05-08 16:52:59 -07:00
|
|
|
|
2021-03-17 22:54:56 -07:00
|
|
|
|
|
|
|
|
|
2021-04-14 11:26:53 -07:00
|
|
|
|
2021-03-17 22:54:56 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-05-08 16:52:59 -07:00
|
|
|
|
2021-04-16 14:44:02 -07:00
|
|
|
|
2021-03-17 22:54:56 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2022-12-11 20:41:42 +01:00
|
|
|
|
2022-07-15 16:05:06 +03:00
|
|
|
|
2021-03-17 22:54:56 -07:00
|
|
|
|
2022-07-28 17:28:37 +03:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-03-17 22:54:56 -07:00
|
|
|
|
|
|
|
|
|
2021-09-14 21:58:22 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-06-25 02:26:48 +01:00
|
|
|
|
2021-03-17 22:54:56 -07:00
|
|
|
|
2022-07-06 12:17:04 +03:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-03-17 22:54:56 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-03-21 19:23:12 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
stage2: guidance on how to implement switch expressions
Here's what I think the ZIR should be. AstGen is not yet implemented to
match this, and the main implementation of analyzeSwitch in Sema is not
yet implemented to match it either.
Here are some example byte size reductions from master branch, with the
ZIR memory layout from this commit:
```
switch (foo) {
a => 1,
b => 2,
c => 3,
d => 4,
}
```
184 bytes (master) => 40 bytes (this branch)
```
switch (foo) {
a, b => 1,
c..d, e, f => 2,
g => 3,
else => 4,
}
```
240 bytes (master) => 80 bytes (this branch)
2021-03-28 23:12:26 -07:00
|
|
|
|
2021-03-29 21:59:08 -07:00
|
|
|
|
|
|
|
|
|
2022-07-19 13:19:23 +03:00
|
|
|
|
2023-05-28 04:31:56 +01:00
|
|
|
|
2022-07-17 16:13:25 +03:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-03-31 21:36:32 -07:00
|
|
|
|
|
|
|
|
|
2022-07-19 20:20:57 +03:00
|
|
|
|
|
|
|
|
|
2021-04-23 18:28:46 -07:00
|
|
|
|
2021-05-04 14:40:59 -07:00
|
|
|
|
2021-10-12 21:26:59 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2022-07-06 11:36:39 +03:00
|
|
|
|
2022-07-12 16:08:23 +03:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2022-07-19 01:25:10 +03:00
|
|
|
|
2022-07-22 17:15:15 +03:00
|
|
|
|
2022-07-28 14:58:20 +03:00
|
|
|
|
2022-10-18 13:45:06 +03:00
|
|
|
|
|
|
|
|
|
2023-02-18 12:26:22 -07:00
|
|
|
|
2023-02-18 14:10:56 -07:00
|
|
|
|
2021-03-17 22:54:56 -07:00
|
|
|
|
2023-05-08 16:52:59 -07:00
|
|
|
|
2021-04-16 14:44:02 -07:00
|
|
|
|
2021-03-17 22:54:56 -07:00
|
|
|
|
|
|
|
|
|
2023-07-18 17:12:19 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-03-17 22:54:56 -07:00
|
|
|
|
|
|
|
|
|
2021-01-16 22:51:01 -07:00
|
|
|
|
|
|
|
|
|
2021-07-14 12:16:48 -07:00
|
|
|
|
stage2 generics improvements: anytype and param type exprs
AstGen result locations now have a `coerced_ty` tag which is the same as
`ty` except it assumes that Sema will do a coercion, so it does not
redundantly add an `as` instruction into the ZIR code. This results in
cleaner ZIR and about a 14% reduction of ZIR bytes.
param and param_comptime ZIR instructions now have a block body for
their type expressions. This allows Sema to skip evaluation of the
block in the case that the parameter is comptime-provided. It also
allows a new mechanism to function: when evaluating type expressions of
generic functions, if it would depend on another parameter, it returns
`error.GenericPoison` which bubbles up and then is caught by the
param/param_comptime instruction and then handled.
This allows parameters to be evaluated independently so that the type
info for functions which have comptime or anytype parameters will still
have types populated for parameters that do not depend on values of
previous parameters (because evaluation of their param blocks will return
successfully instead of `error.GenericPoison`).
It also makes iteration over the block that contains function parameters
slightly more efficient since it now only contains the param
instructions.
Finally, it fixes the case where a generic function type expression contains
a function prototype. Formerly, this situation would cause shared state
to clobber each other; now it is in a proper tree structure so that
can't happen. This fix also required adding a field to Sema
`comptime_args_fn_inst` to make sure that the `comptime_args` field
passed into Sema is applied to the correct `func` instruction.
Source location for `node_offset_asm_ret_ty` is fixed; it was pointing at
the asm output name rather than the return type as intended.
Generic function instantiation is fixed, notably with respect to
parameter type expressions that depend on previous parameters, and with
respect to types which must be always comptime-known. This involves
passing all the comptime arguments at a callsite of a generic function,
and allowing the generic function semantic analysis to coerce the values
to the proper types (since it has access to the evaluated parameter type
expressions) and then decide based on the type whether the parameter is
runtime known or not. In the case of explicitly marked `comptime`
parameters, there is a check at the semantic analysis of the `call`
instruction.
Semantic analysis of `call` instructions does type coercion on the
arguments, which is needed both for generic functions and to make up for
using `coerced_ty` result locations (mentioned above).
Tasks left in this branch:
* Implement the memoization table.
* Add test coverage.
* Improve error reporting and source locations for compile errors.
2021-08-04 21:11:31 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-09-14 21:58:22 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2022-01-17 15:21:58 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
stage2 generics improvements: anytype and param type exprs
AstGen result locations now have a `coerced_ty` tag which is the same as
`ty` except it assumes that Sema will do a coercion, so it does not
redundantly add an `as` instruction into the ZIR code. This results in
cleaner ZIR and about a 14% reduction of ZIR bytes.
param and param_comptime ZIR instructions now have a block body for
their type expressions. This allows Sema to skip evaluation of the
block in the case that the parameter is comptime-provided. It also
allows a new mechanism to function: when evaluating type expressions of
generic functions, if it would depend on another parameter, it returns
`error.GenericPoison` which bubbles up and then is caught by the
param/param_comptime instruction and then handled.
This allows parameters to be evaluated independently so that the type
info for functions which have comptime or anytype parameters will still
have types populated for parameters that do not depend on values of
previous parameters (because evaluation of their param blocks will return
successfully instead of `error.GenericPoison`).
It also makes iteration over the block that contains function parameters
slightly more efficient since it now only contains the param
instructions.
Finally, it fixes the case where a generic function type expression contains
a function prototype. Formerly, this situation would cause shared state
to clobber each other; now it is in a proper tree structure so that
can't happen. This fix also required adding a field to Sema
`comptime_args_fn_inst` to make sure that the `comptime_args` field
passed into Sema is applied to the correct `func` instruction.
Source location for `node_offset_asm_ret_ty` is fixed; it was pointing at
the asm output name rather than the return type as intended.
Generic function instantiation is fixed, notably with respect to
parameter type expressions that depend on previous parameters, and with
respect to types which must be always comptime-known. This involves
passing all the comptime arguments at a callsite of a generic function,
and allowing the generic function semantic analysis to coerce the values
to the proper types (since it has access to the evaluated parameter type
expressions) and then decide based on the type whether the parameter is
runtime known or not. In the case of explicitly marked `comptime`
parameters, there is a check at the semantic analysis of the `call`
instruction.
Semantic analysis of `call` instructions does type coercion on the
arguments, which is needed both for generic functions and to make up for
using `coerced_ty` result locations (mentioned above).
Tasks left in this branch:
* Implement the memoization table.
* Add test coverage.
* Improve error reporting and source locations for compile errors.
2021-08-04 21:11:31 -07:00
|
|
|
|
2020-09-13 19:17:58 -07:00
|
|
|
|
2023-05-02 15:01:45 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
compiler: eliminate Decl.value_arena and Sema.perm_arena
The main motivation for this commit is eliminating Decl.value_arena.
Everything else is dominoes.
Decl.name used to be stored in the GPA, now it is stored in InternPool.
It ended up being simpler to migrate other strings to be interned as
well, such as struct field names, union field names, and a few others.
This ended up requiring a big diff, sorry about that. But the changes
are pretty nice, we finally start to take advantage of InternPool's
existence.
global_error_set and error_name_list are simplified. Now it is a single
ArrayHashMap(NullTerminatedString, void) and the index is the error tag
value.
Module.tmp_hack_arena is re-introduced (it was removed in
eeff407941560ce8eb5b737b2436dfa93cfd3a0c) in order to deal with
comptime_args, optimized_order, and struct and union fields. After
structs and unions get moved into InternPool properly, tmp_hack_arena
can be deleted again.
2023-06-01 23:46:04 -07:00
|
|
|
|
2023-05-02 15:01:45 -07:00
|
|
|
|
|
|
|
|
|
stage2: *WIP*: rework ZIR memory layout; overhaul source locations
The memory layout for ZIR instructions is completely reworked. See
zir.zig for those changes. Some new types:
* `zir.Code`: a "finished" set of ZIR instructions. Instead of allocating
each instruction independently, there is now a Tag and 8 bytes of
data available for all ZIR instructions. Small instructions fit
within these 8 bytes; larger ones use 4 bytes for an index into
`extra`. There is also `string_bytes` so that we can have 4 byte
references to strings. `zir.Inst.Tag` describes how to interpret
those 8 bytes of data.
- This is shared by all `Block` scopes.
* `Module.WipZirCode`: represents an in-progress `zir.Code`. In this
structure, the arrays are mutable, and get resized as we add/delete
things. There is extra state to keep track of things. This struct is
stored on the stack. Once it is finished, it produces an immutable
`zir.Code`, which will remain on the heap for the duration of a
function's existence.
- This is shared by all `GenZir` scopes.
* `Sema`: represents in-progress semantic analysis of a `zir.Code`.
This data is stored on the stack and is shared among all `Block`
scopes. It is now the main "self" argument to everything in the file
that was previously named `zir_sema.zig`.
Additionally, I moved some logic that was in `Module` into here.
`Module.Fn` now stores its parameter names inside the `zir.Code`,
instead of inside ZIR instructions. When the TZIR memory layout
reworking time comes, codegen will be able to reference this data
directly instead of duplicating it.
astgen.zig is (so far) almost entirely untouched, but nearly all of it
will need to be reworked to adhere to this new memory layout structure.
I have no benchmarks to report yet, as I am still working through
compile errors and fixing various things that I broke in this branch.
Overhaul of Source Locations:
Previously we used `usize` everywhere to mean byte offset, but sometimes
also mean other stuff. This was error prone and also made us do
unnecessary work, and store unnecessary bytes in memory.
Now there are more types involved into source locations, and more ways
to describe a source location.
* AllErrors.Message: embrace the assumption that files always have less
than 2 << 32 bytes.
* SrcLoc gets more complicated, to model more complicated source
locations.
* Introduce LazySrcLoc, which can model interesting source locations
with very little stored state. Useful for avoiding doing unnecessary
work when no compile errors occur.
Also, previously, we had `src: usize` on every ZIR instruction. This is
no longer the case. Each instruction now determines whether it even cares
about source location, and if so, how that source location is stored.
This requires more careful work inside `Sema`, but it results in fewer
bytes stored on the heap, without compromising accuracy and power of
compile error messages.
Miscellaneous:
* std.zig: string literals have more helpful result values for
reporting errors. There is now a lower level API and a higher level
API.
- side note: I noticed that the string literal logic needs some love.
There is some unnecessarily hacky code there.
* cut & pasted some TZIR logic that was in zir.zig to ir.zig. This
probably broke stuff and needs to get fixed.
* Removed type/Enum.zig, type/Union.zig, and type/Struct.zig. I don't
think this quite how this code will be organized. Need some more
careful planning about how to implement structs, unions, enums. They
need to be independent Decls, just like a top level function.
2021-03-15 23:38:38 -07:00
|
|
|
|
|
|
|
|
|
2020-09-13 19:17:58 -07:00
|
|
|
|
2021-06-03 15:39:26 -05:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2022-10-24 22:42:01 +03:00
|
|
|
|
|
|
|
|
|
2021-06-03 15:39:26 -05:00
|
|
|
|
|
|
|
|
|
stage2: fix deletion of Decls that get re-referenced
When scanDecls happens, we create stub Decl objects that
have not been semantically analyzed. When they get referenced,
they get semantically analyzed.
Before this commit, when they got unreferenced, they were completely
deleted, including deleted from the containing Namespace.
However, if the update did not cause the containing Namespace to get
deleted, for example, if `std.builtin.ExportOptions` is no longer
referenced, but `std.builtin` is still referenced, and then `ExportOptions`
gets referenced again, the Namespace would be incorrectly missing the
Decl, so we get an incorrect "no such member" error.
The solution is to, when dealing with a no longer referenced Decl
objects during an update, clear them to the state they would be in
on a fresh scanDecl, rather than completely deleting them.
2021-05-18 12:35:36 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-11-24 16:58:01 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
stage2: fix deletion of Decls that get re-referenced
When scanDecls happens, we create stub Decl objects that
have not been semantically analyzed. When they get referenced,
they get semantically analyzed.
Before this commit, when they got unreferenced, they were completely
deleted, including deleted from the containing Namespace.
However, if the update did not cause the containing Namespace to get
deleted, for example, if `std.builtin.ExportOptions` is no longer
referenced, but `std.builtin` is still referenced, and then `ExportOptions`
gets referenced again, the Namespace would be incorrectly missing the
Decl, so we get an incorrect "no such member" error.
The solution is to, when dealing with a no longer referenced Decl
objects during an update, clear them to the state they would be in
on a fresh scanDecl, rather than completely deleting them.
2021-05-18 12:35:36 -07:00
|
|
|
|
stage2: *WIP*: rework ZIR memory layout; overhaul source locations
The memory layout for ZIR instructions is completely reworked. See
zir.zig for those changes. Some new types:
* `zir.Code`: a "finished" set of ZIR instructions. Instead of allocating
each instruction independently, there is now a Tag and 8 bytes of
data available for all ZIR instructions. Small instructions fit
within these 8 bytes; larger ones use 4 bytes for an index into
`extra`. There is also `string_bytes` so that we can have 4 byte
references to strings. `zir.Inst.Tag` describes how to interpret
those 8 bytes of data.
- This is shared by all `Block` scopes.
* `Module.WipZirCode`: represents an in-progress `zir.Code`. In this
structure, the arrays are mutable, and get resized as we add/delete
things. There is extra state to keep track of things. This struct is
stored on the stack. Once it is finished, it produces an immutable
`zir.Code`, which will remain on the heap for the duration of a
function's existence.
- This is shared by all `GenZir` scopes.
* `Sema`: represents in-progress semantic analysis of a `zir.Code`.
This data is stored on the stack and is shared among all `Block`
scopes. It is now the main "self" argument to everything in the file
that was previously named `zir_sema.zig`.
Additionally, I moved some logic that was in `Module` into here.
`Module.Fn` now stores its parameter names inside the `zir.Code`,
instead of inside ZIR instructions. When the TZIR memory layout
reworking time comes, codegen will be able to reference this data
directly instead of duplicating it.
astgen.zig is (so far) almost entirely untouched, but nearly all of it
will need to be reworked to adhere to this new memory layout structure.
I have no benchmarks to report yet, as I am still working through
compile errors and fixing various things that I broke in this branch.
Overhaul of Source Locations:
Previously we used `usize` everywhere to mean byte offset, but sometimes
also mean other stuff. This was error prone and also made us do
unnecessary work, and store unnecessary bytes in memory.
Now there are more types involved into source locations, and more ways
to describe a source location.
* AllErrors.Message: embrace the assumption that files always have less
than 2 << 32 bytes.
* SrcLoc gets more complicated, to model more complicated source
locations.
* Introduce LazySrcLoc, which can model interesting source locations
with very little stored state. Useful for avoiding doing unnecessary
work when no compile errors occur.
Also, previously, we had `src: usize` on every ZIR instruction. This is
no longer the case. Each instruction now determines whether it even cares
about source location, and if so, how that source location is stored.
This requires more careful work inside `Sema`, but it results in fewer
bytes stored on the heap, without compromising accuracy and power of
compile error messages.
Miscellaneous:
* std.zig: string literals have more helpful result values for
reporting errors. There is now a lower level API and a higher level
API.
- side note: I noticed that the string literal logic needs some love.
There is some unnecessarily hacky code there.
* cut & pasted some TZIR logic that was in zir.zig to ir.zig. This
probably broke stuff and needs to get fixed.
* Removed type/Enum.zig, type/Union.zig, and type/Struct.zig. I don't
think this quite how this code will be organized. Need some more
careful planning about how to implement structs, unions, enums. They
need to be independent Decls, just like a top level function.
2021-03-15 23:38:38 -07:00
|
|
|
|
2021-01-16 22:51:01 -07:00
|
|
|
|
stage2: *WIP*: rework ZIR memory layout; overhaul source locations
The memory layout for ZIR instructions is completely reworked. See
zir.zig for those changes. Some new types:
* `zir.Code`: a "finished" set of ZIR instructions. Instead of allocating
each instruction independently, there is now a Tag and 8 bytes of
data available for all ZIR instructions. Small instructions fit
within these 8 bytes; larger ones use 4 bytes for an index into
`extra`. There is also `string_bytes` so that we can have 4 byte
references to strings. `zir.Inst.Tag` describes how to interpret
those 8 bytes of data.
- This is shared by all `Block` scopes.
* `Module.WipZirCode`: represents an in-progress `zir.Code`. In this
structure, the arrays are mutable, and get resized as we add/delete
things. There is extra state to keep track of things. This struct is
stored on the stack. Once it is finished, it produces an immutable
`zir.Code`, which will remain on the heap for the duration of a
function's existence.
- This is shared by all `GenZir` scopes.
* `Sema`: represents in-progress semantic analysis of a `zir.Code`.
This data is stored on the stack and is shared among all `Block`
scopes. It is now the main "self" argument to everything in the file
that was previously named `zir_sema.zig`.
Additionally, I moved some logic that was in `Module` into here.
`Module.Fn` now stores its parameter names inside the `zir.Code`,
instead of inside ZIR instructions. When the TZIR memory layout
reworking time comes, codegen will be able to reference this data
directly instead of duplicating it.
astgen.zig is (so far) almost entirely untouched, but nearly all of it
will need to be reworked to adhere to this new memory layout structure.
I have no benchmarks to report yet, as I am still working through
compile errors and fixing various things that I broke in this branch.
Overhaul of Source Locations:
Previously we used `usize` everywhere to mean byte offset, but sometimes
also mean other stuff. This was error prone and also made us do
unnecessary work, and store unnecessary bytes in memory.
Now there are more types involved into source locations, and more ways
to describe a source location.
* AllErrors.Message: embrace the assumption that files always have less
than 2 << 32 bytes.
* SrcLoc gets more complicated, to model more complicated source
locations.
* Introduce LazySrcLoc, which can model interesting source locations
with very little stored state. Useful for avoiding doing unnecessary
work when no compile errors occur.
Also, previously, we had `src: usize` on every ZIR instruction. This is
no longer the case. Each instruction now determines whether it even cares
about source location, and if so, how that source location is stored.
This requires more careful work inside `Sema`, but it results in fewer
bytes stored on the heap, without compromising accuracy and power of
compile error messages.
Miscellaneous:
* std.zig: string literals have more helpful result values for
reporting errors. There is now a lower level API and a higher level
API.
- side note: I noticed that the string literal logic needs some love.
There is some unnecessarily hacky code there.
* cut & pasted some TZIR logic that was in zir.zig to ir.zig. This
probably broke stuff and needs to get fixed.
* Removed type/Enum.zig, type/Union.zig, and type/Struct.zig. I don't
think this quite how this code will be organized. Need some more
careful planning about how to implement structs, unions, enums. They
need to be independent Decls, just like a top level function.
2021-03-15 23:38:38 -07:00
|
|
|
|
2021-04-25 10:43:07 -07:00
|
|
|
|
|
|
|
|
|
2020-09-13 19:17:58 -07:00
|
|
|
|
2022-10-24 22:42:01 +03:00
|
|
|
|
2021-06-03 15:39:26 -05:00
|
|
|
|
2020-09-13 19:17:58 -07:00
|
|
|
|
2022-10-24 22:42:01 +03:00
|
|
|
|
2020-09-13 19:17:58 -07:00
|
|
|
|
2021-04-26 20:41:07 -07:00
|
|
|
|
2021-06-03 15:39:26 -05:00
|
|
|
|
|
|
|
|
|
2021-04-26 20:41:07 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2022-04-19 21:51:08 -07:00
|
|
|
|
2021-04-26 20:41:07 -07:00
|
|
|
|
2021-01-05 17:33:31 -07:00
|
|
|
|
|
|
|
|
|
2021-06-03 15:39:26 -05:00
|
|
|
|
|
|
|
|
|
2020-09-13 19:17:58 -07:00
|
|
|
|
stage2: *WIP*: rework ZIR memory layout; overhaul source locations
The memory layout for ZIR instructions is completely reworked. See
zir.zig for those changes. Some new types:
* `zir.Code`: a "finished" set of ZIR instructions. Instead of allocating
each instruction independently, there is now a Tag and 8 bytes of
data available for all ZIR instructions. Small instructions fit
within these 8 bytes; larger ones use 4 bytes for an index into
`extra`. There is also `string_bytes` so that we can have 4 byte
references to strings. `zir.Inst.Tag` describes how to interpret
those 8 bytes of data.
- This is shared by all `Block` scopes.
* `Module.WipZirCode`: represents an in-progress `zir.Code`. In this
structure, the arrays are mutable, and get resized as we add/delete
things. There is extra state to keep track of things. This struct is
stored on the stack. Once it is finished, it produces an immutable
`zir.Code`, which will remain on the heap for the duration of a
function's existence.
- This is shared by all `GenZir` scopes.
* `Sema`: represents in-progress semantic analysis of a `zir.Code`.
This data is stored on the stack and is shared among all `Block`
scopes. It is now the main "self" argument to everything in the file
that was previously named `zir_sema.zig`.
Additionally, I moved some logic that was in `Module` into here.
`Module.Fn` now stores its parameter names inside the `zir.Code`,
instead of inside ZIR instructions. When the TZIR memory layout
reworking time comes, codegen will be able to reference this data
directly instead of duplicating it.
astgen.zig is (so far) almost entirely untouched, but nearly all of it
will need to be reworked to adhere to this new memory layout structure.
I have no benchmarks to report yet, as I am still working through
compile errors and fixing various things that I broke in this branch.
Overhaul of Source Locations:
Previously we used `usize` everywhere to mean byte offset, but sometimes
also mean other stuff. This was error prone and also made us do
unnecessary work, and store unnecessary bytes in memory.
Now there are more types involved into source locations, and more ways
to describe a source location.
* AllErrors.Message: embrace the assumption that files always have less
than 2 << 32 bytes.
* SrcLoc gets more complicated, to model more complicated source
locations.
* Introduce LazySrcLoc, which can model interesting source locations
with very little stored state. Useful for avoiding doing unnecessary
work when no compile errors occur.
Also, previously, we had `src: usize` on every ZIR instruction. This is
no longer the case. Each instruction now determines whether it even cares
about source location, and if so, how that source location is stored.
This requires more careful work inside `Sema`, but it results in fewer
bytes stored on the heap, without compromising accuracy and power of
compile error messages.
Miscellaneous:
* std.zig: string literals have more helpful result values for
reporting errors. There is now a lower level API and a higher level
API.
- side note: I noticed that the string literal logic needs some love.
There is some unnecessarily hacky code there.
* cut & pasted some TZIR logic that was in zir.zig to ir.zig. This
probably broke stuff and needs to get fixed.
* Removed type/Enum.zig, type/Union.zig, and type/Struct.zig. I don't
think this quite how this code will be organized. Need some more
careful planning about how to implement structs, unions, enums. They
need to be independent Decls, just like a top level function.
2021-03-15 23:38:38 -07:00
|
|
|
|
2020-09-13 19:17:58 -07:00
|
|
|
|
2021-10-17 18:57:54 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-06-03 15:39:26 -05:00
|
|
|
|
|
|
|
|
|
2020-09-13 19:17:58 -07:00
|
|
|
|
stage2: *WIP*: rework ZIR memory layout; overhaul source locations
The memory layout for ZIR instructions is completely reworked. See
zir.zig for those changes. Some new types:
* `zir.Code`: a "finished" set of ZIR instructions. Instead of allocating
each instruction independently, there is now a Tag and 8 bytes of
data available for all ZIR instructions. Small instructions fit
within these 8 bytes; larger ones use 4 bytes for an index into
`extra`. There is also `string_bytes` so that we can have 4 byte
references to strings. `zir.Inst.Tag` describes how to interpret
those 8 bytes of data.
- This is shared by all `Block` scopes.
* `Module.WipZirCode`: represents an in-progress `zir.Code`. In this
structure, the arrays are mutable, and get resized as we add/delete
things. There is extra state to keep track of things. This struct is
stored on the stack. Once it is finished, it produces an immutable
`zir.Code`, which will remain on the heap for the duration of a
function's existence.
- This is shared by all `GenZir` scopes.
* `Sema`: represents in-progress semantic analysis of a `zir.Code`.
This data is stored on the stack and is shared among all `Block`
scopes. It is now the main "self" argument to everything in the file
that was previously named `zir_sema.zig`.
Additionally, I moved some logic that was in `Module` into here.
`Module.Fn` now stores its parameter names inside the `zir.Code`,
instead of inside ZIR instructions. When the TZIR memory layout
reworking time comes, codegen will be able to reference this data
directly instead of duplicating it.
astgen.zig is (so far) almost entirely untouched, but nearly all of it
will need to be reworked to adhere to this new memory layout structure.
I have no benchmarks to report yet, as I am still working through
compile errors and fixing various things that I broke in this branch.
Overhaul of Source Locations:
Previously we used `usize` everywhere to mean byte offset, but sometimes
also mean other stuff. This was error prone and also made us do
unnecessary work, and store unnecessary bytes in memory.
Now there are more types involved into source locations, and more ways
to describe a source location.
* AllErrors.Message: embrace the assumption that files always have less
than 2 << 32 bytes.
* SrcLoc gets more complicated, to model more complicated source
locations.
* Introduce LazySrcLoc, which can model interesting source locations
with very little stored state. Useful for avoiding doing unnecessary
work when no compile errors occur.
Also, previously, we had `src: usize` on every ZIR instruction. This is
no longer the case. Each instruction now determines whether it even cares
about source location, and if so, how that source location is stored.
This requires more careful work inside `Sema`, but it results in fewer
bytes stored on the heap, without compromising accuracy and power of
compile error messages.
Miscellaneous:
* std.zig: string literals have more helpful result values for
reporting errors. There is now a lower level API and a higher level
API.
- side note: I noticed that the string literal logic needs some love.
There is some unnecessarily hacky code there.
* cut & pasted some TZIR logic that was in zir.zig to ir.zig. This
probably broke stuff and needs to get fixed.
* Removed type/Enum.zig, type/Union.zig, and type/Struct.zig. I don't
think this quite how this code will be organized. Need some more
careful planning about how to implement structs, unions, enums. They
need to be independent Decls, just like a top level function.
2021-03-15 23:38:38 -07:00
|
|
|
|
2020-09-13 19:17:58 -07:00
|
|
|
|
2023-09-25 04:54:07 +02:00
|
|
|
|
|
|
|
|
|
2022-12-12 13:16:09 +02:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
stage2: *WIP*: rework ZIR memory layout; overhaul source locations
The memory layout for ZIR instructions is completely reworked. See
zir.zig for those changes. Some new types:
* `zir.Code`: a "finished" set of ZIR instructions. Instead of allocating
each instruction independently, there is now a Tag and 8 bytes of
data available for all ZIR instructions. Small instructions fit
within these 8 bytes; larger ones use 4 bytes for an index into
`extra`. There is also `string_bytes` so that we can have 4 byte
references to strings. `zir.Inst.Tag` describes how to interpret
those 8 bytes of data.
- This is shared by all `Block` scopes.
* `Module.WipZirCode`: represents an in-progress `zir.Code`. In this
structure, the arrays are mutable, and get resized as we add/delete
things. There is extra state to keep track of things. This struct is
stored on the stack. Once it is finished, it produces an immutable
`zir.Code`, which will remain on the heap for the duration of a
function's existence.
- This is shared by all `GenZir` scopes.
* `Sema`: represents in-progress semantic analysis of a `zir.Code`.
This data is stored on the stack and is shared among all `Block`
scopes. It is now the main "self" argument to everything in the file
that was previously named `zir_sema.zig`.
Additionally, I moved some logic that was in `Module` into here.
`Module.Fn` now stores its parameter names inside the `zir.Code`,
instead of inside ZIR instructions. When the TZIR memory layout
reworking time comes, codegen will be able to reference this data
directly instead of duplicating it.
astgen.zig is (so far) almost entirely untouched, but nearly all of it
will need to be reworked to adhere to this new memory layout structure.
I have no benchmarks to report yet, as I am still working through
compile errors and fixing various things that I broke in this branch.
Overhaul of Source Locations:
Previously we used `usize` everywhere to mean byte offset, but sometimes
also mean other stuff. This was error prone and also made us do
unnecessary work, and store unnecessary bytes in memory.
Now there are more types involved into source locations, and more ways
to describe a source location.
* AllErrors.Message: embrace the assumption that files always have less
than 2 << 32 bytes.
* SrcLoc gets more complicated, to model more complicated source
locations.
* Introduce LazySrcLoc, which can model interesting source locations
with very little stored state. Useful for avoiding doing unnecessary
work when no compile errors occur.
Also, previously, we had `src: usize` on every ZIR instruction. This is
no longer the case. Each instruction now determines whether it even cares
about source location, and if so, how that source location is stored.
This requires more careful work inside `Sema`, but it results in fewer
bytes stored on the heap, without compromising accuracy and power of
compile error messages.
Miscellaneous:
* std.zig: string literals have more helpful result values for
reporting errors. There is now a lower level API and a higher level
API.
- side note: I noticed that the string literal logic needs some love.
There is some unnecessarily hacky code there.
* cut & pasted some TZIR logic that was in zir.zig to ir.zig. This
probably broke stuff and needs to get fixed.
* Removed type/Enum.zig, type/Union.zig, and type/Struct.zig. I don't
think this quite how this code will be organized. Need some more
careful planning about how to implement structs, unions, enums. They
need to be independent Decls, just like a top level function.
2021-03-15 23:38:38 -07:00
|
|
|
|
2020-11-21 21:12:33 -05:00
|
|
|
|
2022-11-27 01:07:35 -07:00
|
|
|
|
|
|
|
|
|
2020-09-13 19:17:58 -07:00
|
|
|
|
stage2: *WIP*: rework ZIR memory layout; overhaul source locations
The memory layout for ZIR instructions is completely reworked. See
zir.zig for those changes. Some new types:
* `zir.Code`: a "finished" set of ZIR instructions. Instead of allocating
each instruction independently, there is now a Tag and 8 bytes of
data available for all ZIR instructions. Small instructions fit
within these 8 bytes; larger ones use 4 bytes for an index into
`extra`. There is also `string_bytes` so that we can have 4 byte
references to strings. `zir.Inst.Tag` describes how to interpret
those 8 bytes of data.
- This is shared by all `Block` scopes.
* `Module.WipZirCode`: represents an in-progress `zir.Code`. In this
structure, the arrays are mutable, and get resized as we add/delete
things. There is extra state to keep track of things. This struct is
stored on the stack. Once it is finished, it produces an immutable
`zir.Code`, which will remain on the heap for the duration of a
function's existence.
- This is shared by all `GenZir` scopes.
* `Sema`: represents in-progress semantic analysis of a `zir.Code`.
This data is stored on the stack and is shared among all `Block`
scopes. It is now the main "self" argument to everything in the file
that was previously named `zir_sema.zig`.
Additionally, I moved some logic that was in `Module` into here.
`Module.Fn` now stores its parameter names inside the `zir.Code`,
instead of inside ZIR instructions. When the TZIR memory layout
reworking time comes, codegen will be able to reference this data
directly instead of duplicating it.
astgen.zig is (so far) almost entirely untouched, but nearly all of it
will need to be reworked to adhere to this new memory layout structure.
I have no benchmarks to report yet, as I am still working through
compile errors and fixing various things that I broke in this branch.
Overhaul of Source Locations:
Previously we used `usize` everywhere to mean byte offset, but sometimes
also mean other stuff. This was error prone and also made us do
unnecessary work, and store unnecessary bytes in memory.
Now there are more types involved into source locations, and more ways
to describe a source location.
* AllErrors.Message: embrace the assumption that files always have less
than 2 << 32 bytes.
* SrcLoc gets more complicated, to model more complicated source
locations.
* Introduce LazySrcLoc, which can model interesting source locations
with very little stored state. Useful for avoiding doing unnecessary
work when no compile errors occur.
Also, previously, we had `src: usize` on every ZIR instruction. This is
no longer the case. Each instruction now determines whether it even cares
about source location, and if so, how that source location is stored.
This requires more careful work inside `Sema`, but it results in fewer
bytes stored on the heap, without compromising accuracy and power of
compile error messages.
Miscellaneous:
* std.zig: string literals have more helpful result values for
reporting errors. There is now a lower level API and a higher level
API.
- side note: I noticed that the string literal logic needs some love.
There is some unnecessarily hacky code there.
* cut & pasted some TZIR logic that was in zir.zig to ir.zig. This
probably broke stuff and needs to get fixed.
* Removed type/Enum.zig, type/Union.zig, and type/Struct.zig. I don't
think this quite how this code will be organized. Need some more
careful planning about how to implement structs, unions, enums. They
need to be independent Decls, just like a top level function.
2021-03-15 23:38:38 -07:00
|
|
|
|
2020-09-13 19:17:58 -07:00
|
|
|
|
2022-11-27 01:07:35 -07:00
|
|
|
|
2021-06-03 15:39:26 -05:00
|
|
|
|
2020-09-13 19:17:58 -07:00
|
|
|
|
stage2: *WIP*: rework ZIR memory layout; overhaul source locations
The memory layout for ZIR instructions is completely reworked. See
zir.zig for those changes. Some new types:
* `zir.Code`: a "finished" set of ZIR instructions. Instead of allocating
each instruction independently, there is now a Tag and 8 bytes of
data available for all ZIR instructions. Small instructions fit
within these 8 bytes; larger ones use 4 bytes for an index into
`extra`. There is also `string_bytes` so that we can have 4 byte
references to strings. `zir.Inst.Tag` describes how to interpret
those 8 bytes of data.
- This is shared by all `Block` scopes.
* `Module.WipZirCode`: represents an in-progress `zir.Code`. In this
structure, the arrays are mutable, and get resized as we add/delete
things. There is extra state to keep track of things. This struct is
stored on the stack. Once it is finished, it produces an immutable
`zir.Code`, which will remain on the heap for the duration of a
function's existence.
- This is shared by all `GenZir` scopes.
* `Sema`: represents in-progress semantic analysis of a `zir.Code`.
This data is stored on the stack and is shared among all `Block`
scopes. It is now the main "self" argument to everything in the file
that was previously named `zir_sema.zig`.
Additionally, I moved some logic that was in `Module` into here.
`Module.Fn` now stores its parameter names inside the `zir.Code`,
instead of inside ZIR instructions. When the TZIR memory layout
reworking time comes, codegen will be able to reference this data
directly instead of duplicating it.
astgen.zig is (so far) almost entirely untouched, but nearly all of it
will need to be reworked to adhere to this new memory layout structure.
I have no benchmarks to report yet, as I am still working through
compile errors and fixing various things that I broke in this branch.
Overhaul of Source Locations:
Previously we used `usize` everywhere to mean byte offset, but sometimes
also mean other stuff. This was error prone and also made us do
unnecessary work, and store unnecessary bytes in memory.
Now there are more types involved into source locations, and more ways
to describe a source location.
* AllErrors.Message: embrace the assumption that files always have less
than 2 << 32 bytes.
* SrcLoc gets more complicated, to model more complicated source
locations.
* Introduce LazySrcLoc, which can model interesting source locations
with very little stored state. Useful for avoiding doing unnecessary
work when no compile errors occur.
Also, previously, we had `src: usize` on every ZIR instruction. This is
no longer the case. Each instruction now determines whether it even cares
about source location, and if so, how that source location is stored.
This requires more careful work inside `Sema`, but it results in fewer
bytes stored on the heap, without compromising accuracy and power of
compile error messages.
Miscellaneous:
* std.zig: string literals have more helpful result values for
reporting errors. There is now a lower level API and a higher level
API.
- side note: I noticed that the string literal logic needs some love.
There is some unnecessarily hacky code there.
* cut & pasted some TZIR logic that was in zir.zig to ir.zig. This
probably broke stuff and needs to get fixed.
* Removed type/Enum.zig, type/Union.zig, and type/Struct.zig. I don't
think this quite how this code will be organized. Need some more
careful planning about how to implement structs, unions, enums. They
need to be independent Decls, just like a top level function.
2021-03-15 23:38:38 -07:00
|
|
|
|
2020-09-13 19:17:58 -07:00
|
|
|
|
compiler: eliminate Decl.value_arena and Sema.perm_arena
The main motivation for this commit is eliminating Decl.value_arena.
Everything else is dominoes.
Decl.name used to be stored in the GPA, now it is stored in InternPool.
It ended up being simpler to migrate other strings to be interned as
well, such as struct field names, union field names, and a few others.
This ended up requiring a big diff, sorry about that. But the changes
are pretty nice, we finally start to take advantage of InternPool's
existence.
global_error_set and error_name_list are simplified. Now it is a single
ArrayHashMap(NullTerminatedString, void) and the index is the error tag
value.
Module.tmp_hack_arena is re-introduced (it was removed in
eeff407941560ce8eb5b737b2436dfa93cfd3a0c) in order to deal with
comptime_args, optimized_order, and struct and union fields. After
structs and unions get moved into InternPool properly, tmp_hack_arena
can be deleted again.
2023-06-01 23:46:04 -07:00
|
|
|
|
2020-09-09 18:21:50 +03:00
|
|
|
|
2021-07-27 15:44:21 -07:00
|
|
|
|
2022-04-19 21:51:08 -07:00
|
|
|
|
2022-05-04 20:38:53 -07:00
|
|
|
|
2022-09-09 18:55:58 +03:00
|
|
|
|
2022-05-24 00:10:56 -07:00
|
|
|
|
2023-05-02 15:01:45 -07:00
|
|
|
|
compiler: eliminate Decl.value_arena and Sema.perm_arena
The main motivation for this commit is eliminating Decl.value_arena.
Everything else is dominoes.
Decl.name used to be stored in the GPA, now it is stored in InternPool.
It ended up being simpler to migrate other strings to be interned as
well, such as struct field names, union field names, and a few others.
This ended up requiring a big diff, sorry about that. But the changes
are pretty nice, we finally start to take advantage of InternPool's
existence.
global_error_set and error_name_list are simplified. Now it is a single
ArrayHashMap(NullTerminatedString, void) and the index is the error tag
value.
Module.tmp_hack_arena is re-introduced (it was removed in
eeff407941560ce8eb5b737b2436dfa93cfd3a0c) in order to deal with
comptime_args, optimized_order, and struct and union fields. After
structs and unions get moved into InternPool properly, tmp_hack_arena
can be deleted again.
2023-06-01 23:46:04 -07:00
|
|
|
|
2023-09-14 19:59:58 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2022-04-19 21:51:08 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-07-06 15:27:47 -07:00
|
|
|
|
|
|
|
|
|
2022-04-19 21:51:08 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2022-05-04 20:38:53 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2022-04-19 21:51:08 -07:00
|
|
|
|
2023-05-26 16:04:53 -04:00
|
|
|
|
2023-05-08 16:52:59 -07:00
|
|
|
|
|
|
|
|
|
2022-04-19 21:51:08 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-07-06 15:27:47 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2022-04-19 21:51:08 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-05-08 16:52:59 -07:00
|
|
|
|
2023-07-06 15:27:47 -07:00
|
|
|
|
2023-05-08 16:52:59 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-07-06 15:27:47 -07:00
|
|
|
|
2022-04-19 21:51:08 -07:00
|
|
|
|
|
|
|
|
|
2023-05-26 16:04:53 -04:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2022-04-19 21:51:08 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-05-08 16:52:59 -07:00
|
|
|
|
|
|
|
|
|
2022-04-19 21:51:08 -07:00
|
|
|
|
2023-05-08 16:52:59 -07:00
|
|
|
|
2020-09-13 19:17:58 -07:00
|
|
|
|
|
|
|
|
|
2022-11-27 01:07:35 -07:00
|
|
|
|
compiler: eliminate Decl.value_arena and Sema.perm_arena
The main motivation for this commit is eliminating Decl.value_arena.
Everything else is dominoes.
Decl.name used to be stored in the GPA, now it is stored in InternPool.
It ended up being simpler to migrate other strings to be interned as
well, such as struct field names, union field names, and a few others.
This ended up requiring a big diff, sorry about that. But the changes
are pretty nice, we finally start to take advantage of InternPool's
existence.
global_error_set and error_name_list are simplified. Now it is a single
ArrayHashMap(NullTerminatedString, void) and the index is the error tag
value.
Module.tmp_hack_arena is re-introduced (it was removed in
eeff407941560ce8eb5b737b2436dfa93cfd3a0c) in order to deal with
comptime_args, optimized_order, and struct and union fields. After
structs and unions get moved into InternPool properly, tmp_hack_arena
can be deleted again.
2023-06-01 23:46:04 -07:00
|
|
|
|
2022-11-27 01:07:35 -07:00
|
|
|
|
2020-09-13 19:17:58 -07:00
|
|
|
|
|
|
|
|
|
2022-12-06 20:35:50 -07:00
|
|
|
|
2021-04-25 00:02:58 -07:00
|
|
|
|
2022-12-06 20:35:50 -07:00
|
|
|
|
2021-09-15 21:05:40 +02:00
|
|
|
|
2021-04-26 20:41:07 -07:00
|
|
|
|
2021-04-25 00:02:58 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2022-12-06 20:35:50 -07:00
|
|
|
|
2021-04-25 00:02:58 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-10-01 17:45:13 -05:00
|
|
|
|
2021-04-15 20:34:21 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-04-14 11:26:53 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-10-06 21:29:08 -07:00
|
|
|
|
2021-04-25 00:02:58 -07:00
|
|
|
|
2021-04-14 11:26:53 -07:00
|
|
|
|
2021-04-25 00:02:58 -07:00
|
|
|
|
|
|
|
|
|
2023-10-06 21:29:08 -07:00
|
|
|
|
2021-04-25 00:02:58 -07:00
|
|
|
|
|
|
|
|
|
2021-07-03 11:47:58 -07:00
|
|
|
|
2023-03-05 12:46:12 -07:00
|
|
|
|
2021-04-25 00:02:58 -07:00
|
|
|
|
2023-10-06 21:29:08 -07:00
|
|
|
|
|
|
|
|
|
2021-04-25 00:02:58 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-04-25 10:43:07 -07:00
|
|
|
|
|
|
|
|
|
2021-04-25 00:02:58 -07:00
|
|
|
|
2021-04-14 11:26:53 -07:00
|
|
|
|
2023-03-06 17:22:23 -05:00
|
|
|
|
|
|
|
|
|
2021-04-25 00:02:58 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-05-20 23:11:53 +01:00
|
|
|
|
2023-03-06 17:22:23 -05:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-04-25 00:02:58 -07:00
|
|
|
|
2023-03-06 17:22:23 -05:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-04-25 00:02:58 -07:00
|
|
|
|
2023-03-06 17:22:23 -05:00
|
|
|
|
2021-04-25 00:02:58 -07:00
|
|
|
|
2023-05-20 23:11:53 +01:00
|
|
|
|
2023-03-06 17:22:23 -05:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2022-07-16 15:46:13 +03:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-03-06 17:22:23 -05:00
|
|
|
|
2022-07-16 15:46:13 +03:00
|
|
|
|
|
|
|
|
|
2023-03-06 17:22:23 -05:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-04-25 00:02:58 -07:00
|
|
|
|
2023-03-06 17:22:23 -05:00
|
|
|
|
2021-04-25 00:02:58 -07:00
|
|
|
|
|
|
|
|
|
2023-03-06 17:22:23 -05:00
|
|
|
|
2021-04-25 00:02:58 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-03-06 17:22:23 -05:00
|
|
|
|
2021-04-25 00:02:58 -07:00
|
|
|
|
2021-04-26 20:41:07 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-04-25 00:02:58 -07:00
|
|
|
|
2023-05-05 03:04:44 -04:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-04-25 00:02:58 -07:00
|
|
|
|
|
|
|
|
|
2021-12-30 20:49:02 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-04-28 22:43:26 -07:00
|
|
|
|
2021-04-25 00:02:58 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-11-09 18:27:12 -07:00
|
|
|
|
|
|
|
|
|
2021-04-25 00:02:58 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-03-06 17:22:23 -05:00
|
|
|
|
2021-04-14 11:26:53 -07:00
|
|
|
|
2023-03-06 17:22:23 -05:00
|
|
|
|
2023-05-20 23:11:53 +01:00
|
|
|
|
2023-03-06 17:22:23 -05:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-05-20 23:11:53 +01:00
|
|
|
|
2023-03-06 17:22:23 -05:00
|
|
|
|
2021-04-14 11:26:53 -07:00
|
|
|
|
2021-04-25 00:02:58 -07:00
|
|
|
|
2023-03-06 17:22:23 -05:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-04-25 00:02:58 -07:00
|
|
|
|
|
|
|
|
|
2021-04-28 16:55:22 -07:00
|
|
|
|
|
|
|
|
|
2021-05-12 22:02:44 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-04-14 11:26:53 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-06-22 18:46:56 +01:00
|
|
|
|
2021-04-14 11:26:53 -07:00
|
|
|
|
2021-04-25 00:02:58 -07:00
|
|
|
|
2021-04-14 11:26:53 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-12-30 20:49:02 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-04-14 11:26:53 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-02-02 17:32:06 -07:00
|
|
|
|
2021-04-14 11:26:53 -07:00
|
|
|
|
|
|
|
|
|
2023-02-23 16:18:43 -07:00
|
|
|
|
2021-05-02 17:08:19 -07:00
|
|
|
|
2021-04-14 11:26:53 -07:00
|
|
|
|
2021-04-28 22:43:26 -07:00
|
|
|
|
2021-04-25 00:02:58 -07:00
|
|
|
|
2021-04-14 11:26:53 -07:00
|
|
|
|
2021-04-25 00:02:58 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-08-28 15:15:46 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-06-22 18:46:56 +01:00
|
|
|
|
2021-04-25 00:02:58 -07:00
|
|
|
|
2023-06-22 18:46:56 +01:00
|
|
|
|
2021-04-25 00:02:58 -07:00
|
|
|
|
|
|
|
|
|
2023-02-18 09:02:57 -07:00
|
|
|
|
2023-06-22 18:46:56 +01:00
|
|
|
|
2021-04-25 00:02:58 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-06-22 18:46:56 +01:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-04-25 00:02:58 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-06-22 18:46:56 +01:00
|
|
|
|
2021-04-25 00:02:58 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-06-22 18:46:56 +01:00
|
|
|
|
2021-04-25 00:02:58 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-06-22 18:46:56 +01:00
|
|
|
|
2021-04-25 00:02:58 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-03-06 17:22:23 -05:00
|
|
|
|
2023-10-06 21:29:08 -07:00
|
|
|
|
|
|
|
|
|
2021-04-25 00:02:58 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-04-16 14:44:02 -07:00
|
|
|
|
2021-04-14 11:26:53 -07:00
|
|
|
|
2021-11-09 18:27:12 -07:00
|
|
|
|
|
|
|
|
|
2021-04-16 14:44:02 -07:00
|
|
|
|
2021-04-14 11:26:53 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-05-12 22:02:44 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2022-04-19 21:51:08 -07:00
|
|
|
|
2021-05-12 22:02:44 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2022-04-19 21:51:08 -07:00
|
|
|
|
2021-05-12 22:02:44 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-04-14 11:26:53 -07:00
|
|
|
|
|
|
|
|
|
2023-05-05 03:04:44 -04:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-06-22 18:46:56 +01:00
|
|
|
|
2023-05-05 03:04:44 -04:00
|
|
|
|
2023-06-22 18:46:56 +01:00
|
|
|
|
2023-05-05 03:04:44 -04:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-06-22 18:46:56 +01:00
|
|
|
|
2023-05-05 03:04:44 -04:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-06-22 18:46:56 +01:00
|
|
|
|
2023-05-05 03:04:44 -04:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-06-15 13:14:16 +06:00
|
|
|
|
2023-06-22 18:46:56 +01:00
|
|
|
|
2023-05-05 03:04:44 -04:00
|
|
|
|
2023-06-15 13:14:16 +06:00
|
|
|
|
2023-05-05 03:04:44 -04:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-05-05 13:16:14 -07:00
|
|
|
|
|
|
|
|
|
2021-05-07 14:18:14 -07:00
|
|
|
|
2021-05-05 16:56:24 -07:00
|
|
|
|
2021-05-05 13:16:14 -07:00
|
|
|
|
2022-04-19 21:51:08 -07:00
|
|
|
|
|
|
|
|
|
2021-05-05 13:16:14 -07:00
|
|
|
|
|
|
|
|
|
2023-03-13 18:23:53 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-05-05 13:16:14 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-05-06 17:20:45 -07:00
|
|
|
|
|
|
|
|
|
2021-05-05 13:16:14 -07:00
|
|
|
|
2022-11-27 01:07:35 -07:00
|
|
|
|
2021-05-05 13:16:14 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-05-06 17:20:45 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-05-05 13:16:14 -07:00
|
|
|
|
rework generic function calls
Abridged summary:
* Move `Module.Fn` into `InternPool`.
* Delete a lot of confusing and problematic `Sema` logic related to
generic function calls.
This commit removes `Module.Fn` and replaces it with two new
`InternPool.Tag` values:
* `func_decl` - corresponding to a function declared in the source
code. This one contains line/column numbers, zir_body_inst, etc.
* `func_instance` - one for each monomorphization of a generic
function. Contains a reference to the `func_decl` from whence the
instantiation came, along with the `comptime` parameter values (or
types in the case of `anytype`)
Since `InternPool` provides deduplication on these values, these fields
are now deleted from `Module`:
* `monomorphed_func_keys`
* `monomorphed_funcs`
* `align_stack_fns`
Instead of these, Sema logic for generic function instantiation now
unconditionally evaluates the function prototype expression for every
generic callsite. This is technically required in order for type
coercions to work. The previous code had some dubious, probably wrong
hacks to make things work, such as `hashUncoerced`. I'm not 100% sure
how we were able to eliminate that function and still pass all the
behavior tests, but I'm pretty sure things were still broken without
doing type coercion for every generic function call argument.
After the function prototype is evaluated, it produces a deduplicated
`func_instance` `InternPool.Index` which can then be used for the
generic function call.
Some other nice things made by this simplification are the removal of
`comptime_args_fn_inst` and `preallocated_new_func` from `Sema`, and the
messy logic associated with them.
I have not yet been able to measure the perf of this against master
branch. On one hand, it reduces memory usage and pointer chasing of the
most heavily used `InternPool` Tag - function bodies - but on the other
hand, it does evaluate function prototype expressions more than before.
We will soon find out.
2023-07-03 22:09:30 -07:00
|
|
|
|
|
|
|
|
|
2022-04-19 21:51:08 -07:00
|
|
|
|
|
|
|
|
|
2021-05-05 13:16:14 -07:00
|
|
|
|
|
|
|
|
|
2021-05-06 17:20:45 -07:00
|
|
|
|
|
|
|
|
|
2021-05-05 13:16:14 -07:00
|
|
|
|
2021-05-12 22:02:44 -07:00
|
|
|
|
|
|
|
|
|
2022-04-19 21:51:08 -07:00
|
|
|
|
2021-05-05 13:16:14 -07:00
|
|
|
|
|
|
|
|
|
2021-05-12 22:02:44 -07:00
|
|
|
|
|
|
|
|
|
2021-05-05 13:16:14 -07:00
|
|
|
|
|
|
|
|
|
2022-04-19 21:51:08 -07:00
|
|
|
|
2021-05-05 13:16:14 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-05-11 22:12:36 -07:00
|
|
|
|
2021-05-05 13:16:14 -07:00
|
|
|
|
2023-08-24 20:43:43 -07:00
|
|
|
|
|
|
|
|
|
2022-04-19 21:51:08 -07:00
|
|
|
|
2021-05-05 13:16:14 -07:00
|
|
|
|
2023-08-24 20:43:43 -07:00
|
|
|
|
2021-05-05 13:16:14 -07:00
|
|
|
|
|
|
|
|
|
2023-08-21 14:27:34 -07:00
|
|
|
|
|
|
|
|
|
2022-04-19 21:51:08 -07:00
|
|
|
|
2021-05-07 14:18:14 -07:00
|
|
|
|
2023-08-21 14:27:34 -07:00
|
|
|
|
2021-05-07 14:18:14 -07:00
|
|
|
|
|
|
|
|
|
2023-05-26 16:04:53 -04:00
|
|
|
|
rework generic function calls
Abridged summary:
* Move `Module.Fn` into `InternPool`.
* Delete a lot of confusing and problematic `Sema` logic related to
generic function calls.
This commit removes `Module.Fn` and replaces it with two new
`InternPool.Tag` values:
* `func_decl` - corresponding to a function declared in the source
code. This one contains line/column numbers, zir_body_inst, etc.
* `func_instance` - one for each monomorphization of a generic
function. Contains a reference to the `func_decl` from whence the
instantiation came, along with the `comptime` parameter values (or
types in the case of `anytype`)
Since `InternPool` provides deduplication on these values, these fields
are now deleted from `Module`:
* `monomorphed_func_keys`
* `monomorphed_funcs`
* `align_stack_fns`
Instead of these, Sema logic for generic function instantiation now
unconditionally evaluates the function prototype expression for every
generic callsite. This is technically required in order for type
coercions to work. The previous code had some dubious, probably wrong
hacks to make things work, such as `hashUncoerced`. I'm not 100% sure
how we were able to eliminate that function and still pass all the
behavior tests, but I'm pretty sure things were still broken without
doing type coercion for every generic function call argument.
After the function prototype is evaluated, it produces a deduplicated
`func_instance` `InternPool.Index` which can then be used for the
generic function call.
Some other nice things made by this simplification are the removal of
`comptime_args_fn_inst` and `preallocated_new_func` from `Sema`, and the
messy logic associated with them.
I have not yet been able to measure the perf of this against master
branch. On one hand, it reduces memory usage and pointer chasing of the
most heavily used `InternPool` Tag - function bodies - but on the other
hand, it does evaluate function prototype expressions more than before.
We will soon find out.
2023-07-03 22:09:30 -07:00
|
|
|
|
2022-04-19 21:51:08 -07:00
|
|
|
|
2021-05-05 16:56:24 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-05-26 16:04:53 -04:00
|
|
|
|
2022-03-25 22:53:58 +08:00
|
|
|
|
2021-05-05 13:16:14 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
stage2: improve handling of the generated file builtin.zig
All Zig code is eligible to `@import("builtin")` which is mapped to a
generated file, build.zig, based on the target and other settings.
Zig invocations which share the same target settings will generate the
same builtin.zig file and thus the path to builtin.zig is in a shared
cache folder, and different projects can sometimes use the same file.
Before this commit, this led to race conditions where multiple
invocations of `zig` would race to write this file. If one process
wanted to *read* the file while the other process *wrote* the file, the
reading process could observe a truncated or partially written
builtin.zig file.
This commit makes the following improvements:
- limitations:
- avoid clobbering the inode, mtime in the hot path
- avoid creating a partially written file
- builtin.zig needs to be on disk for debug info / stack trace purposes
- don't mark the task as complete until the file is finished being populated
(possibly by an external process)
- strategy:
- create the `@import("builtin")` `Module.File` during the AstGen
work, based on generating the contents in memory rather than
loading from disk.
- write builtin.zig in a separate task that doesn't have
to complete until the end of the AstGen work queue so that it
can be done in parallel with everything else.
- when writing the file, first stat the file path. If it exists, we are done.
- otherwise, write the file to a temp file in the same directory and atomically
rename it into place (clobbering the inode, mtime in the cold path).
- summary:
- all limitations respected
- hot path: one stat() syscall that happens in a worker thread
This required adding a missing function to the standard library:
`std.fs.Dir.statFile`. In this commit, it does open() and then fstat()
which is two syscalls. It should be improved in a future commit to only
make one.
Fixes #9439.
2021-12-06 16:38:12 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-10-06 21:29:08 -07:00
|
|
|
|
stage2: improve handling of the generated file builtin.zig
All Zig code is eligible to `@import("builtin")` which is mapped to a
generated file, build.zig, based on the target and other settings.
Zig invocations which share the same target settings will generate the
same builtin.zig file and thus the path to builtin.zig is in a shared
cache folder, and different projects can sometimes use the same file.
Before this commit, this led to race conditions where multiple
invocations of `zig` would race to write this file. If one process
wanted to *read* the file while the other process *wrote* the file, the
reading process could observe a truncated or partially written
builtin.zig file.
This commit makes the following improvements:
- limitations:
- avoid clobbering the inode, mtime in the hot path
- avoid creating a partially written file
- builtin.zig needs to be on disk for debug info / stack trace purposes
- don't mark the task as complete until the file is finished being populated
(possibly by an external process)
- strategy:
- create the `@import("builtin")` `Module.File` during the AstGen
work, based on generating the contents in memory rather than
loading from disk.
- write builtin.zig in a separate task that doesn't have
to complete until the end of the AstGen work queue so that it
can be done in parallel with everything else.
- when writing the file, first stat the file path. If it exists, we are done.
- otherwise, write the file to a temp file in the same directory and atomically
rename it into place (clobbering the inode, mtime in the cold path).
- summary:
- all limitations respected
- hot path: one stat() syscall that happens in a worker thread
This required adding a missing function to the standard library:
`std.fs.Dir.statFile`. In this commit, it does open() and then fstat()
which is two syscalls. It should be improved in a future commit to only
make one.
Fixes #9439.
2021-12-06 16:38:12 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-10-06 21:29:08 -07:00
|
|
|
|
2023-10-06 17:05:58 -07:00
|
|
|
|
2023-10-06 21:29:08 -07:00
|
|
|
|
stage2: improve handling of the generated file builtin.zig
All Zig code is eligible to `@import("builtin")` which is mapped to a
generated file, build.zig, based on the target and other settings.
Zig invocations which share the same target settings will generate the
same builtin.zig file and thus the path to builtin.zig is in a shared
cache folder, and different projects can sometimes use the same file.
Before this commit, this led to race conditions where multiple
invocations of `zig` would race to write this file. If one process
wanted to *read* the file while the other process *wrote* the file, the
reading process could observe a truncated or partially written
builtin.zig file.
This commit makes the following improvements:
- limitations:
- avoid clobbering the inode, mtime in the hot path
- avoid creating a partially written file
- builtin.zig needs to be on disk for debug info / stack trace purposes
- don't mark the task as complete until the file is finished being populated
(possibly by an external process)
- strategy:
- create the `@import("builtin")` `Module.File` during the AstGen
work, based on generating the contents in memory rather than
loading from disk.
- write builtin.zig in a separate task that doesn't have
to complete until the end of the AstGen work queue so that it
can be done in parallel with everything else.
- when writing the file, first stat the file path. If it exists, we are done.
- otherwise, write the file to a temp file in the same directory and atomically
rename it into place (clobbering the inode, mtime in the cold path).
- summary:
- all limitations respected
- hot path: one stat() syscall that happens in a worker thread
This required adding a missing function to the standard library:
`std.fs.Dir.statFile`. In this commit, it does open() and then fstat()
which is two syscalls. It should be improved in a future commit to only
make one.
Fixes #9439.
2021-12-06 16:38:12 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-10-06 21:29:08 -07:00
|
|
|
|
stage2: improve handling of the generated file builtin.zig
All Zig code is eligible to `@import("builtin")` which is mapped to a
generated file, build.zig, based on the target and other settings.
Zig invocations which share the same target settings will generate the
same builtin.zig file and thus the path to builtin.zig is in a shared
cache folder, and different projects can sometimes use the same file.
Before this commit, this led to race conditions where multiple
invocations of `zig` would race to write this file. If one process
wanted to *read* the file while the other process *wrote* the file, the
reading process could observe a truncated or partially written
builtin.zig file.
This commit makes the following improvements:
- limitations:
- avoid clobbering the inode, mtime in the hot path
- avoid creating a partially written file
- builtin.zig needs to be on disk for debug info / stack trace purposes
- don't mark the task as complete until the file is finished being populated
(possibly by an external process)
- strategy:
- create the `@import("builtin")` `Module.File` during the AstGen
work, based on generating the contents in memory rather than
loading from disk.
- write builtin.zig in a separate task that doesn't have
to complete until the end of the AstGen work queue so that it
can be done in parallel with everything else.
- when writing the file, first stat the file path. If it exists, we are done.
- otherwise, write the file to a temp file in the same directory and atomically
rename it into place (clobbering the inode, mtime in the cold path).
- summary:
- all limitations respected
- hot path: one stat() syscall that happens in a worker thread
This required adding a missing function to the standard library:
`std.fs.Dir.statFile`. In this commit, it does open() and then fstat()
which is two syscalls. It should be improved in a future commit to only
make one.
Fixes #9439.
2021-12-06 16:38:12 -07:00
|
|
|
|
|
|
|
|
|
2023-10-06 21:29:08 -07:00
|
|
|
|
stage2: improve handling of the generated file builtin.zig
All Zig code is eligible to `@import("builtin")` which is mapped to a
generated file, build.zig, based on the target and other settings.
Zig invocations which share the same target settings will generate the
same builtin.zig file and thus the path to builtin.zig is in a shared
cache folder, and different projects can sometimes use the same file.
Before this commit, this led to race conditions where multiple
invocations of `zig` would race to write this file. If one process
wanted to *read* the file while the other process *wrote* the file, the
reading process could observe a truncated or partially written
builtin.zig file.
This commit makes the following improvements:
- limitations:
- avoid clobbering the inode, mtime in the hot path
- avoid creating a partially written file
- builtin.zig needs to be on disk for debug info / stack trace purposes
- don't mark the task as complete until the file is finished being populated
(possibly by an external process)
- strategy:
- create the `@import("builtin")` `Module.File` during the AstGen
work, based on generating the contents in memory rather than
loading from disk.
- write builtin.zig in a separate task that doesn't have
to complete until the end of the AstGen work queue so that it
can be done in parallel with everything else.
- when writing the file, first stat the file path. If it exists, we are done.
- otherwise, write the file to a temp file in the same directory and atomically
rename it into place (clobbering the inode, mtime in the cold path).
- summary:
- all limitations respected
- hot path: one stat() syscall that happens in a worker thread
This required adding a missing function to the standard library:
`std.fs.Dir.statFile`. In this commit, it does open() and then fstat()
which is two syscalls. It should be improved in a future commit to only
make one.
Fixes #9439.
2021-12-06 16:38:12 -07:00
|
|
|
|
2023-10-06 21:29:08 -07:00
|
|
|
|
stage2: improve handling of the generated file builtin.zig
All Zig code is eligible to `@import("builtin")` which is mapped to a
generated file, build.zig, based on the target and other settings.
Zig invocations which share the same target settings will generate the
same builtin.zig file and thus the path to builtin.zig is in a shared
cache folder, and different projects can sometimes use the same file.
Before this commit, this led to race conditions where multiple
invocations of `zig` would race to write this file. If one process
wanted to *read* the file while the other process *wrote* the file, the
reading process could observe a truncated or partially written
builtin.zig file.
This commit makes the following improvements:
- limitations:
- avoid clobbering the inode, mtime in the hot path
- avoid creating a partially written file
- builtin.zig needs to be on disk for debug info / stack trace purposes
- don't mark the task as complete until the file is finished being populated
(possibly by an external process)
- strategy:
- create the `@import("builtin")` `Module.File` during the AstGen
work, based on generating the contents in memory rather than
loading from disk.
- write builtin.zig in a separate task that doesn't have
to complete until the end of the AstGen work queue so that it
can be done in parallel with everything else.
- when writing the file, first stat the file path. If it exists, we are done.
- otherwise, write the file to a temp file in the same directory and atomically
rename it into place (clobbering the inode, mtime in the cold path).
- summary:
- all limitations respected
- hot path: one stat() syscall that happens in a worker thread
This required adding a missing function to the standard library:
`std.fs.Dir.statFile`. In this commit, it does open() and then fstat()
which is two syscalls. It should be improved in a future commit to only
make one.
Fixes #9439.
2021-12-06 16:38:12 -07:00
|
|
|
|
|
|
|
|
|
2023-10-06 17:05:58 -07:00
|
|
|
|
stage2: improve handling of the generated file builtin.zig
All Zig code is eligible to `@import("builtin")` which is mapped to a
generated file, build.zig, based on the target and other settings.
Zig invocations which share the same target settings will generate the
same builtin.zig file and thus the path to builtin.zig is in a shared
cache folder, and different projects can sometimes use the same file.
Before this commit, this led to race conditions where multiple
invocations of `zig` would race to write this file. If one process
wanted to *read* the file while the other process *wrote* the file, the
reading process could observe a truncated or partially written
builtin.zig file.
This commit makes the following improvements:
- limitations:
- avoid clobbering the inode, mtime in the hot path
- avoid creating a partially written file
- builtin.zig needs to be on disk for debug info / stack trace purposes
- don't mark the task as complete until the file is finished being populated
(possibly by an external process)
- strategy:
- create the `@import("builtin")` `Module.File` during the AstGen
work, based on generating the contents in memory rather than
loading from disk.
- write builtin.zig in a separate task that doesn't have
to complete until the end of the AstGen work queue so that it
can be done in parallel with everything else.
- when writing the file, first stat the file path. If it exists, we are done.
- otherwise, write the file to a temp file in the same directory and atomically
rename it into place (clobbering the inode, mtime in the cold path).
- summary:
- all limitations respected
- hot path: one stat() syscall that happens in a worker thread
This required adding a missing function to the standard library:
`std.fs.Dir.statFile`. In this commit, it does open() and then fstat()
which is two syscalls. It should be improved in a future commit to only
make one.
Fixes #9439.
2021-12-06 16:38:12 -07:00
|
|
|
|
2021-12-30 20:49:02 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
stage2: improve handling of the generated file builtin.zig
All Zig code is eligible to `@import("builtin")` which is mapped to a
generated file, build.zig, based on the target and other settings.
Zig invocations which share the same target settings will generate the
same builtin.zig file and thus the path to builtin.zig is in a shared
cache folder, and different projects can sometimes use the same file.
Before this commit, this led to race conditions where multiple
invocations of `zig` would race to write this file. If one process
wanted to *read* the file while the other process *wrote* the file, the
reading process could observe a truncated or partially written
builtin.zig file.
This commit makes the following improvements:
- limitations:
- avoid clobbering the inode, mtime in the hot path
- avoid creating a partially written file
- builtin.zig needs to be on disk for debug info / stack trace purposes
- don't mark the task as complete until the file is finished being populated
(possibly by an external process)
- strategy:
- create the `@import("builtin")` `Module.File` during the AstGen
work, based on generating the contents in memory rather than
loading from disk.
- write builtin.zig in a separate task that doesn't have
to complete until the end of the AstGen work queue so that it
can be done in parallel with everything else.
- when writing the file, first stat the file path. If it exists, we are done.
- otherwise, write the file to a temp file in the same directory and atomically
rename it into place (clobbering the inode, mtime in the cold path).
- summary:
- all limitations respected
- hot path: one stat() syscall that happens in a worker thread
This required adding a missing function to the standard library:
`std.fs.Dir.statFile`. In this commit, it does open() and then fstat()
which is two syscalls. It should be improved in a future commit to only
make one.
Fixes #9439.
2021-12-06 16:38:12 -07:00
|
|
|
|
|
|
|
|
|
2022-09-09 09:55:09 -07:00
|
|
|
|
stage2: improve handling of the generated file builtin.zig
All Zig code is eligible to `@import("builtin")` which is mapped to a
generated file, build.zig, based on the target and other settings.
Zig invocations which share the same target settings will generate the
same builtin.zig file and thus the path to builtin.zig is in a shared
cache folder, and different projects can sometimes use the same file.
Before this commit, this led to race conditions where multiple
invocations of `zig` would race to write this file. If one process
wanted to *read* the file while the other process *wrote* the file, the
reading process could observe a truncated or partially written
builtin.zig file.
This commit makes the following improvements:
- limitations:
- avoid clobbering the inode, mtime in the hot path
- avoid creating a partially written file
- builtin.zig needs to be on disk for debug info / stack trace purposes
- don't mark the task as complete until the file is finished being populated
(possibly by an external process)
- strategy:
- create the `@import("builtin")` `Module.File` during the AstGen
work, based on generating the contents in memory rather than
loading from disk.
- write builtin.zig in a separate task that doesn't have
to complete until the end of the AstGen work queue so that it
can be done in parallel with everything else.
- when writing the file, first stat the file path. If it exists, we are done.
- otherwise, write the file to a temp file in the same directory and atomically
rename it into place (clobbering the inode, mtime in the cold path).
- summary:
- all limitations respected
- hot path: one stat() syscall that happens in a worker thread
This required adding a missing function to the standard library:
`std.fs.Dir.statFile`. In this commit, it does open() and then fstat()
which is two syscalls. It should be improved in a future commit to only
make one.
Fixes #9439.
2021-12-06 16:38:12 -07:00
|
|
|
|
2022-09-09 09:55:09 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-10-06 17:05:58 -07:00
|
|
|
|
2022-09-09 09:55:09 -07:00
|
|
|
|
|
|
|
|
|
stage2: improve handling of the generated file builtin.zig
All Zig code is eligible to `@import("builtin")` which is mapped to a
generated file, build.zig, based on the target and other settings.
Zig invocations which share the same target settings will generate the
same builtin.zig file and thus the path to builtin.zig is in a shared
cache folder, and different projects can sometimes use the same file.
Before this commit, this led to race conditions where multiple
invocations of `zig` would race to write this file. If one process
wanted to *read* the file while the other process *wrote* the file, the
reading process could observe a truncated or partially written
builtin.zig file.
This commit makes the following improvements:
- limitations:
- avoid clobbering the inode, mtime in the hot path
- avoid creating a partially written file
- builtin.zig needs to be on disk for debug info / stack trace purposes
- don't mark the task as complete until the file is finished being populated
(possibly by an external process)
- strategy:
- create the `@import("builtin")` `Module.File` during the AstGen
work, based on generating the contents in memory rather than
loading from disk.
- write builtin.zig in a separate task that doesn't have
to complete until the end of the AstGen work queue so that it
can be done in parallel with everything else.
- when writing the file, first stat the file path. If it exists, we are done.
- otherwise, write the file to a temp file in the same directory and atomically
rename it into place (clobbering the inode, mtime in the cold path).
- summary:
- all limitations respected
- hot path: one stat() syscall that happens in a worker thread
This required adding a missing function to the standard library:
`std.fs.Dir.statFile`. In this commit, it does open() and then fstat()
which is two syscalls. It should be improved in a future commit to only
make one.
Fixes #9439.
2021-12-06 16:38:12 -07:00
|
|
|
|
|
|
|
|
|
2023-02-02 17:32:06 -07:00
|
|
|
|
stage2: improve handling of the generated file builtin.zig
All Zig code is eligible to `@import("builtin")` which is mapped to a
generated file, build.zig, based on the target and other settings.
Zig invocations which share the same target settings will generate the
same builtin.zig file and thus the path to builtin.zig is in a shared
cache folder, and different projects can sometimes use the same file.
Before this commit, this led to race conditions where multiple
invocations of `zig` would race to write this file. If one process
wanted to *read* the file while the other process *wrote* the file, the
reading process could observe a truncated or partially written
builtin.zig file.
This commit makes the following improvements:
- limitations:
- avoid clobbering the inode, mtime in the hot path
- avoid creating a partially written file
- builtin.zig needs to be on disk for debug info / stack trace purposes
- don't mark the task as complete until the file is finished being populated
(possibly by an external process)
- strategy:
- create the `@import("builtin")` `Module.File` during the AstGen
work, based on generating the contents in memory rather than
loading from disk.
- write builtin.zig in a separate task that doesn't have
to complete until the end of the AstGen work queue so that it
can be done in parallel with everything else.
- when writing the file, first stat the file path. If it exists, we are done.
- otherwise, write the file to a temp file in the same directory and atomically
rename it into place (clobbering the inode, mtime in the cold path).
- summary:
- all limitations respected
- hot path: one stat() syscall that happens in a worker thread
This required adding a missing function to the standard library:
`std.fs.Dir.statFile`. In this commit, it does open() and then fstat()
which is two syscalls. It should be improved in a future commit to only
make one.
Fixes #9439.
2021-12-06 16:38:12 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-10-06 17:05:58 -07:00
|
|
|
|
2023-10-06 21:29:08 -07:00
|
|
|
|
stage2: improve handling of the generated file builtin.zig
All Zig code is eligible to `@import("builtin")` which is mapped to a
generated file, build.zig, based on the target and other settings.
Zig invocations which share the same target settings will generate the
same builtin.zig file and thus the path to builtin.zig is in a shared
cache folder, and different projects can sometimes use the same file.
Before this commit, this led to race conditions where multiple
invocations of `zig` would race to write this file. If one process
wanted to *read* the file while the other process *wrote* the file, the
reading process could observe a truncated or partially written
builtin.zig file.
This commit makes the following improvements:
- limitations:
- avoid clobbering the inode, mtime in the hot path
- avoid creating a partially written file
- builtin.zig needs to be on disk for debug info / stack trace purposes
- don't mark the task as complete until the file is finished being populated
(possibly by an external process)
- strategy:
- create the `@import("builtin")` `Module.File` during the AstGen
work, based on generating the contents in memory rather than
loading from disk.
- write builtin.zig in a separate task that doesn't have
to complete until the end of the AstGen work queue so that it
can be done in parallel with everything else.
- when writing the file, first stat the file path. If it exists, we are done.
- otherwise, write the file to a temp file in the same directory and atomically
rename it into place (clobbering the inode, mtime in the cold path).
- summary:
- all limitations respected
- hot path: one stat() syscall that happens in a worker thread
This required adding a missing function to the standard library:
`std.fs.Dir.statFile`. In this commit, it does open() and then fstat()
which is two syscalls. It should be improved in a future commit to only
make one.
Fixes #9439.
2021-12-06 16:38:12 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-12-30 20:49:02 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
stage2: improve handling of the generated file builtin.zig
All Zig code is eligible to `@import("builtin")` which is mapped to a
generated file, build.zig, based on the target and other settings.
Zig invocations which share the same target settings will generate the
same builtin.zig file and thus the path to builtin.zig is in a shared
cache folder, and different projects can sometimes use the same file.
Before this commit, this led to race conditions where multiple
invocations of `zig` would race to write this file. If one process
wanted to *read* the file while the other process *wrote* the file, the
reading process could observe a truncated or partially written
builtin.zig file.
This commit makes the following improvements:
- limitations:
- avoid clobbering the inode, mtime in the hot path
- avoid creating a partially written file
- builtin.zig needs to be on disk for debug info / stack trace purposes
- don't mark the task as complete until the file is finished being populated
(possibly by an external process)
- strategy:
- create the `@import("builtin")` `Module.File` during the AstGen
work, based on generating the contents in memory rather than
loading from disk.
- write builtin.zig in a separate task that doesn't have
to complete until the end of the AstGen work queue so that it
can be done in parallel with everything else.
- when writing the file, first stat the file path. If it exists, we are done.
- otherwise, write the file to a temp file in the same directory and atomically
rename it into place (clobbering the inode, mtime in the cold path).
- summary:
- all limitations respected
- hot path: one stat() syscall that happens in a worker thread
This required adding a missing function to the standard library:
`std.fs.Dir.statFile`. In this commit, it does open() and then fstat()
which is two syscalls. It should be improved in a future commit to only
make one.
Fixes #9439.
2021-12-06 16:38:12 -07:00
|
|
|
|
|
|
|
|
|
2021-05-05 13:16:14 -07:00
|
|
|
|
2021-10-29 00:37:25 +01:00
|
|
|
|
2021-05-05 13:16:14 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2022-11-27 01:07:35 -07:00
|
|
|
|
2021-05-05 13:16:14 -07:00
|
|
|
|
|
|
|
|
|
2021-09-23 12:17:06 -05:00
|
|
|
|
2021-05-05 13:16:14 -07:00
|
|
|
|
2021-09-23 12:17:06 -05:00
|
|
|
|
|
|
|
|
|
2021-05-05 13:16:14 -07:00
|
|
|
|
|
|
|
|
|
2021-05-05 16:56:24 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-05-05 13:16:14 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-05-05 16:56:24 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-05-05 13:16:14 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2022-03-11 19:31:29 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2022-04-19 21:51:08 -07:00
|
|
|
|
2020-09-13 19:17:58 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2022-04-19 21:51:08 -07:00
|
|
|
|
|
|
|
|
|
2020-09-13 19:17:58 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-05-11 17:34:13 -07:00
|
|
|
|
2020-09-13 19:17:58 -07:00
|
|
|
|
|
|
|
|
|
2023-04-10 00:00:35 -04:00
|
|
|
|
2020-09-13 19:17:58 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-01-30 18:22:50 +01:00
|
|
|
|
2022-08-10 20:39:57 -07:00
|
|
|
|
2020-09-13 19:17:58 -07:00
|
|
|
|
2022-04-19 21:51:08 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-05-14 17:41:22 -07:00
|
|
|
|
2022-04-19 21:51:08 -07:00
|
|
|
|
2020-09-13 19:17:58 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-06-02 00:33:28 -07:00
|
|
|
|
2022-03-31 19:07:05 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-04-19 17:40:29 +01:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-05-08 16:52:59 -07:00
|
|
|
|
2023-04-19 17:40:29 +01:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-01-16 22:51:01 -07:00
|
|
|
|
2020-09-13 19:17:58 -07:00
|
|
|
|
|
|
|
|
|
2023-04-19 17:40:29 +01:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-05-02 15:01:45 -07:00
|
|
|
|
2020-09-13 19:17:58 -07:00
|
|
|
|
2023-04-19 17:40:29 +01:00
|
|
|
|
2023-06-15 13:14:16 +06:00
|
|
|
|
2023-04-19 17:40:29 +01:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-04-10 00:00:35 -04:00
|
|
|
|
2023-04-19 17:40:29 +01:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2020-09-13 19:17:58 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
rework generic function calls
Abridged summary:
* Move `Module.Fn` into `InternPool`.
* Delete a lot of confusing and problematic `Sema` logic related to
generic function calls.
This commit removes `Module.Fn` and replaces it with two new
`InternPool.Tag` values:
* `func_decl` - corresponding to a function declared in the source
code. This one contains line/column numbers, zir_body_inst, etc.
* `func_instance` - one for each monomorphization of a generic
function. Contains a reference to the `func_decl` from whence the
instantiation came, along with the `comptime` parameter values (or
types in the case of `anytype`)
Since `InternPool` provides deduplication on these values, these fields
are now deleted from `Module`:
* `monomorphed_func_keys`
* `monomorphed_funcs`
* `align_stack_fns`
Instead of these, Sema logic for generic function instantiation now
unconditionally evaluates the function prototype expression for every
generic callsite. This is technically required in order for type
coercions to work. The previous code had some dubious, probably wrong
hacks to make things work, such as `hashUncoerced`. I'm not 100% sure
how we were able to eliminate that function and still pass all the
behavior tests, but I'm pretty sure things were still broken without
doing type coercion for every generic function call argument.
After the function prototype is evaluated, it produces a deduplicated
`func_instance` `InternPool.Index` which can then be used for the
generic function call.
Some other nice things made by this simplification are the removal of
`comptime_args_fn_inst` and `preallocated_new_func` from `Sema`, and the
messy logic associated with them.
I have not yet been able to measure the perf of this against master
branch. On one hand, it reduces memory usage and pointer chasing of the
most heavily used `InternPool` Tag - function bodies - but on the other
hand, it does evaluate function prototype expressions more than before.
We will soon find out.
2023-07-03 22:09:30 -07:00
|
|
|
|
2022-03-11 19:31:29 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
rework generic function calls
Abridged summary:
* Move `Module.Fn` into `InternPool`.
* Delete a lot of confusing and problematic `Sema` logic related to
generic function calls.
This commit removes `Module.Fn` and replaces it with two new
`InternPool.Tag` values:
* `func_decl` - corresponding to a function declared in the source
code. This one contains line/column numbers, zir_body_inst, etc.
* `func_instance` - one for each monomorphization of a generic
function. Contains a reference to the `func_decl` from whence the
instantiation came, along with the `comptime` parameter values (or
types in the case of `anytype`)
Since `InternPool` provides deduplication on these values, these fields
are now deleted from `Module`:
* `monomorphed_func_keys`
* `monomorphed_funcs`
* `align_stack_fns`
Instead of these, Sema logic for generic function instantiation now
unconditionally evaluates the function prototype expression for every
generic callsite. This is technically required in order for type
coercions to work. The previous code had some dubious, probably wrong
hacks to make things work, such as `hashUncoerced`. I'm not 100% sure
how we were able to eliminate that function and still pass all the
behavior tests, but I'm pretty sure things were still broken without
doing type coercion for every generic function call argument.
After the function prototype is evaluated, it produces a deduplicated
`func_instance` `InternPool.Index` which can then be used for the
generic function call.
Some other nice things made by this simplification are the removal of
`comptime_args_fn_inst` and `preallocated_new_func` from `Sema`, and the
messy logic associated with them.
I have not yet been able to measure the perf of this against master
branch. On one hand, it reduces memory usage and pointer chasing of the
most heavily used `InternPool` Tag - function bodies - but on the other
hand, it does evaluate function prototype expressions more than before.
We will soon find out.
2023-07-03 22:09:30 -07:00
|
|
|
|
|
|
|
|
|
2022-04-19 21:51:08 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2022-03-11 19:31:29 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-04-10 00:00:35 -04:00
|
|
|
|
2022-03-11 19:31:29 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
rework generic function calls
Abridged summary:
* Move `Module.Fn` into `InternPool`.
* Delete a lot of confusing and problematic `Sema` logic related to
generic function calls.
This commit removes `Module.Fn` and replaces it with two new
`InternPool.Tag` values:
* `func_decl` - corresponding to a function declared in the source
code. This one contains line/column numbers, zir_body_inst, etc.
* `func_instance` - one for each monomorphization of a generic
function. Contains a reference to the `func_decl` from whence the
instantiation came, along with the `comptime` parameter values (or
types in the case of `anytype`)
Since `InternPool` provides deduplication on these values, these fields
are now deleted from `Module`:
* `monomorphed_func_keys`
* `monomorphed_funcs`
* `align_stack_fns`
Instead of these, Sema logic for generic function instantiation now
unconditionally evaluates the function prototype expression for every
generic callsite. This is technically required in order for type
coercions to work. The previous code had some dubious, probably wrong
hacks to make things work, such as `hashUncoerced`. I'm not 100% sure
how we were able to eliminate that function and still pass all the
behavior tests, but I'm pretty sure things were still broken without
doing type coercion for every generic function call argument.
After the function prototype is evaluated, it produces a deduplicated
`func_instance` `InternPool.Index` which can then be used for the
generic function call.
Some other nice things made by this simplification are the removal of
`comptime_args_fn_inst` and `preallocated_new_func` from `Sema`, and the
messy logic associated with them.
I have not yet been able to measure the perf of this against master
branch. On one hand, it reduces memory usage and pointer chasing of the
most heavily used `InternPool` Tag - function bodies - but on the other
hand, it does evaluate function prototype expressions more than before.
We will soon find out.
2023-07-03 22:09:30 -07:00
|
|
|
|
2022-03-11 19:31:29 -07:00
|
|
|
|
2023-05-29 05:07:17 +01:00
|
|
|
|
2022-03-11 19:31:29 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-05-22 07:58:02 -04:00
|
|
|
|
2022-03-11 19:31:29 -07:00
|
|
|
|
rework generic function calls
Abridged summary:
* Move `Module.Fn` into `InternPool`.
* Delete a lot of confusing and problematic `Sema` logic related to
generic function calls.
This commit removes `Module.Fn` and replaces it with two new
`InternPool.Tag` values:
* `func_decl` - corresponding to a function declared in the source
code. This one contains line/column numbers, zir_body_inst, etc.
* `func_instance` - one for each monomorphization of a generic
function. Contains a reference to the `func_decl` from whence the
instantiation came, along with the `comptime` parameter values (or
types in the case of `anytype`)
Since `InternPool` provides deduplication on these values, these fields
are now deleted from `Module`:
* `monomorphed_func_keys`
* `monomorphed_funcs`
* `align_stack_fns`
Instead of these, Sema logic for generic function instantiation now
unconditionally evaluates the function prototype expression for every
generic callsite. This is technically required in order for type
coercions to work. The previous code had some dubious, probably wrong
hacks to make things work, such as `hashUncoerced`. I'm not 100% sure
how we were able to eliminate that function and still pass all the
behavior tests, but I'm pretty sure things were still broken without
doing type coercion for every generic function call argument.
After the function prototype is evaluated, it produces a deduplicated
`func_instance` `InternPool.Index` which can then be used for the
generic function call.
Some other nice things made by this simplification are the removal of
`comptime_args_fn_inst` and `preallocated_new_func` from `Sema`, and the
messy logic associated with them.
I have not yet been able to measure the perf of this against master
branch. On one hand, it reduces memory usage and pointer chasing of the
most heavily used `InternPool` Tag - function bodies - but on the other
hand, it does evaluate function prototype expressions more than before.
We will soon find out.
2023-07-03 22:09:30 -07:00
|
|
|
|
2022-03-11 19:31:29 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
rework generic function calls
Abridged summary:
* Move `Module.Fn` into `InternPool`.
* Delete a lot of confusing and problematic `Sema` logic related to
generic function calls.
This commit removes `Module.Fn` and replaces it with two new
`InternPool.Tag` values:
* `func_decl` - corresponding to a function declared in the source
code. This one contains line/column numbers, zir_body_inst, etc.
* `func_instance` - one for each monomorphization of a generic
function. Contains a reference to the `func_decl` from whence the
instantiation came, along with the `comptime` parameter values (or
types in the case of `anytype`)
Since `InternPool` provides deduplication on these values, these fields
are now deleted from `Module`:
* `monomorphed_func_keys`
* `monomorphed_funcs`
* `align_stack_fns`
Instead of these, Sema logic for generic function instantiation now
unconditionally evaluates the function prototype expression for every
generic callsite. This is technically required in order for type
coercions to work. The previous code had some dubious, probably wrong
hacks to make things work, such as `hashUncoerced`. I'm not 100% sure
how we were able to eliminate that function and still pass all the
behavior tests, but I'm pretty sure things were still broken without
doing type coercion for every generic function call argument.
After the function prototype is evaluated, it produces a deduplicated
`func_instance` `InternPool.Index` which can then be used for the
generic function call.
Some other nice things made by this simplification are the removal of
`comptime_args_fn_inst` and `preallocated_new_func` from `Sema`, and the
messy logic associated with them.
I have not yet been able to measure the perf of this against master
branch. On one hand, it reduces memory usage and pointer chasing of the
most heavily used `InternPool` Tag - function bodies - but on the other
hand, it does evaluate function prototype expressions more than before.
We will soon find out.
2023-07-03 22:09:30 -07:00
|
|
|
|
2022-03-11 19:31:29 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2022-07-24 14:31:00 -07:00
|
|
|
|
|
|
|
|
|
2022-09-06 13:28:31 +02:00
|
|
|
|
2022-07-24 14:31:00 -07:00
|
|
|
|
|
|
|
|
|
2022-09-06 13:28:31 +02:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-03-16 01:32:43 -04:00
|
|
|
|
2022-09-06 13:28:31 +02:00
|
|
|
|
2022-10-15 02:34:34 -04:00
|
|
|
|
2022-03-11 19:31:29 -07:00
|
|
|
|
rework generic function calls
Abridged summary:
* Move `Module.Fn` into `InternPool`.
* Delete a lot of confusing and problematic `Sema` logic related to
generic function calls.
This commit removes `Module.Fn` and replaces it with two new
`InternPool.Tag` values:
* `func_decl` - corresponding to a function declared in the source
code. This one contains line/column numbers, zir_body_inst, etc.
* `func_instance` - one for each monomorphization of a generic
function. Contains a reference to the `func_decl` from whence the
instantiation came, along with the `comptime` parameter values (or
types in the case of `anytype`)
Since `InternPool` provides deduplication on these values, these fields
are now deleted from `Module`:
* `monomorphed_func_keys`
* `monomorphed_funcs`
* `align_stack_fns`
Instead of these, Sema logic for generic function instantiation now
unconditionally evaluates the function prototype expression for every
generic callsite. This is technically required in order for type
coercions to work. The previous code had some dubious, probably wrong
hacks to make things work, such as `hashUncoerced`. I'm not 100% sure
how we were able to eliminate that function and still pass all the
behavior tests, but I'm pretty sure things were still broken without
doing type coercion for every generic function call argument.
After the function prototype is evaluated, it produces a deduplicated
`func_instance` `InternPool.Index` which can then be used for the
generic function call.
Some other nice things made by this simplification are the removal of
`comptime_args_fn_inst` and `preallocated_new_func` from `Sema`, and the
messy logic associated with them.
I have not yet been able to measure the perf of this against master
branch. On one hand, it reduces memory usage and pointer chasing of the
most heavily used `InternPool` Tag - function bodies - but on the other
hand, it does evaluate function prototype expressions more than before.
We will soon find out.
2023-07-03 22:09:30 -07:00
|
|
|
|
2022-03-11 19:31:29 -07:00
|
|
|
|
|
|
|
|
|
2022-09-06 13:28:31 +02:00
|
|
|
|
2023-06-12 01:44:12 -04:00
|
|
|
|
rework generic function calls
Abridged summary:
* Move `Module.Fn` into `InternPool`.
* Delete a lot of confusing and problematic `Sema` logic related to
generic function calls.
This commit removes `Module.Fn` and replaces it with two new
`InternPool.Tag` values:
* `func_decl` - corresponding to a function declared in the source
code. This one contains line/column numbers, zir_body_inst, etc.
* `func_instance` - one for each monomorphization of a generic
function. Contains a reference to the `func_decl` from whence the
instantiation came, along with the `comptime` parameter values (or
types in the case of `anytype`)
Since `InternPool` provides deduplication on these values, these fields
are now deleted from `Module`:
* `monomorphed_func_keys`
* `monomorphed_funcs`
* `align_stack_fns`
Instead of these, Sema logic for generic function instantiation now
unconditionally evaluates the function prototype expression for every
generic callsite. This is technically required in order for type
coercions to work. The previous code had some dubious, probably wrong
hacks to make things work, such as `hashUncoerced`. I'm not 100% sure
how we were able to eliminate that function and still pass all the
behavior tests, but I'm pretty sure things were still broken without
doing type coercion for every generic function call argument.
After the function prototype is evaluated, it produces a deduplicated
`func_instance` `InternPool.Index` which can then be used for the
generic function call.
Some other nice things made by this simplification are the removal of
`comptime_args_fn_inst` and `preallocated_new_func` from `Sema`, and the
messy logic associated with them.
I have not yet been able to measure the perf of this against master
branch. On one hand, it reduces memory usage and pointer chasing of the
most heavily used `InternPool` Tag - function bodies - but on the other
hand, it does evaluate function prototype expressions more than before.
We will soon find out.
2023-07-03 22:09:30 -07:00
|
|
|
|
2022-04-19 21:51:08 -07:00
|
|
|
|
rework generic function calls
Abridged summary:
* Move `Module.Fn` into `InternPool`.
* Delete a lot of confusing and problematic `Sema` logic related to
generic function calls.
This commit removes `Module.Fn` and replaces it with two new
`InternPool.Tag` values:
* `func_decl` - corresponding to a function declared in the source
code. This one contains line/column numbers, zir_body_inst, etc.
* `func_instance` - one for each monomorphization of a generic
function. Contains a reference to the `func_decl` from whence the
instantiation came, along with the `comptime` parameter values (or
types in the case of `anytype`)
Since `InternPool` provides deduplication on these values, these fields
are now deleted from `Module`:
* `monomorphed_func_keys`
* `monomorphed_funcs`
* `align_stack_fns`
Instead of these, Sema logic for generic function instantiation now
unconditionally evaluates the function prototype expression for every
generic callsite. This is technically required in order for type
coercions to work. The previous code had some dubious, probably wrong
hacks to make things work, such as `hashUncoerced`. I'm not 100% sure
how we were able to eliminate that function and still pass all the
behavior tests, but I'm pretty sure things were still broken without
doing type coercion for every generic function call argument.
After the function prototype is evaluated, it produces a deduplicated
`func_instance` `InternPool.Index` which can then be used for the
generic function call.
Some other nice things made by this simplification are the removal of
`comptime_args_fn_inst` and `preallocated_new_func` from `Sema`, and the
messy logic associated with them.
I have not yet been able to measure the perf of this against master
branch. On one hand, it reduces memory usage and pointer chasing of the
most heavily used `InternPool` Tag - function bodies - but on the other
hand, it does evaluate function prototype expressions more than before.
We will soon find out.
2023-07-03 22:09:30 -07:00
|
|
|
|
2022-03-11 19:31:29 -07:00
|
|
|
|
|
|
|
|
|
2023-04-10 00:00:35 -04:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
rework generic function calls
Abridged summary:
* Move `Module.Fn` into `InternPool`.
* Delete a lot of confusing and problematic `Sema` logic related to
generic function calls.
This commit removes `Module.Fn` and replaces it with two new
`InternPool.Tag` values:
* `func_decl` - corresponding to a function declared in the source
code. This one contains line/column numbers, zir_body_inst, etc.
* `func_instance` - one for each monomorphization of a generic
function. Contains a reference to the `func_decl` from whence the
instantiation came, along with the `comptime` parameter values (or
types in the case of `anytype`)
Since `InternPool` provides deduplication on these values, these fields
are now deleted from `Module`:
* `monomorphed_func_keys`
* `monomorphed_funcs`
* `align_stack_fns`
Instead of these, Sema logic for generic function instantiation now
unconditionally evaluates the function prototype expression for every
generic callsite. This is technically required in order for type
coercions to work. The previous code had some dubious, probably wrong
hacks to make things work, such as `hashUncoerced`. I'm not 100% sure
how we were able to eliminate that function and still pass all the
behavior tests, but I'm pretty sure things were still broken without
doing type coercion for every generic function call argument.
After the function prototype is evaluated, it produces a deduplicated
`func_instance` `InternPool.Index` which can then be used for the
generic function call.
Some other nice things made by this simplification are the removal of
`comptime_args_fn_inst` and `preallocated_new_func` from `Sema`, and the
messy logic associated with them.
I have not yet been able to measure the perf of this against master
branch. On one hand, it reduces memory usage and pointer chasing of the
most heavily used `InternPool` Tag - function bodies - but on the other
hand, it does evaluate function prototype expressions more than before.
We will soon find out.
2023-07-03 22:09:30 -07:00
|
|
|
|
2023-04-10 00:00:35 -04:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-05-08 16:52:59 -07:00
|
|
|
|
2023-04-10 00:00:35 -04:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2022-10-15 02:34:34 -04:00
|
|
|
|
2022-09-06 13:28:31 +02:00
|
|
|
|
2023-05-22 07:58:02 -04:00
|
|
|
|
2022-03-11 19:31:29 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2022-04-19 21:51:08 -07:00
|
|
|
|
2022-03-11 19:31:29 -07:00
|
|
|
|
2023-05-08 16:52:59 -07:00
|
|
|
|
2022-03-11 19:31:29 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-05-29 05:07:17 +01:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
rework generic function calls
Abridged summary:
* Move `Module.Fn` into `InternPool`.
* Delete a lot of confusing and problematic `Sema` logic related to
generic function calls.
This commit removes `Module.Fn` and replaces it with two new
`InternPool.Tag` values:
* `func_decl` - corresponding to a function declared in the source
code. This one contains line/column numbers, zir_body_inst, etc.
* `func_instance` - one for each monomorphization of a generic
function. Contains a reference to the `func_decl` from whence the
instantiation came, along with the `comptime` parameter values (or
types in the case of `anytype`)
Since `InternPool` provides deduplication on these values, these fields
are now deleted from `Module`:
* `monomorphed_func_keys`
* `monomorphed_funcs`
* `align_stack_fns`
Instead of these, Sema logic for generic function instantiation now
unconditionally evaluates the function prototype expression for every
generic callsite. This is technically required in order for type
coercions to work. The previous code had some dubious, probably wrong
hacks to make things work, such as `hashUncoerced`. I'm not 100% sure
how we were able to eliminate that function and still pass all the
behavior tests, but I'm pretty sure things were still broken without
doing type coercion for every generic function call argument.
After the function prototype is evaluated, it produces a deduplicated
`func_instance` `InternPool.Index` which can then be used for the
generic function call.
Some other nice things made by this simplification are the removal of
`comptime_args_fn_inst` and `preallocated_new_func` from `Sema`, and the
messy logic associated with them.
I have not yet been able to measure the perf of this against master
branch. On one hand, it reduces memory usage and pointer chasing of the
most heavily used `InternPool` Tag - function bodies - but on the other
hand, it does evaluate function prototype expressions more than before.
We will soon find out.
2023-07-03 22:09:30 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-05-29 05:07:17 +01:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
rework generic function calls
Abridged summary:
* Move `Module.Fn` into `InternPool`.
* Delete a lot of confusing and problematic `Sema` logic related to
generic function calls.
This commit removes `Module.Fn` and replaces it with two new
`InternPool.Tag` values:
* `func_decl` - corresponding to a function declared in the source
code. This one contains line/column numbers, zir_body_inst, etc.
* `func_instance` - one for each monomorphization of a generic
function. Contains a reference to the `func_decl` from whence the
instantiation came, along with the `comptime` parameter values (or
types in the case of `anytype`)
Since `InternPool` provides deduplication on these values, these fields
are now deleted from `Module`:
* `monomorphed_func_keys`
* `monomorphed_funcs`
* `align_stack_fns`
Instead of these, Sema logic for generic function instantiation now
unconditionally evaluates the function prototype expression for every
generic callsite. This is technically required in order for type
coercions to work. The previous code had some dubious, probably wrong
hacks to make things work, such as `hashUncoerced`. I'm not 100% sure
how we were able to eliminate that function and still pass all the
behavior tests, but I'm pretty sure things were still broken without
doing type coercion for every generic function call argument.
After the function prototype is evaluated, it produces a deduplicated
`func_instance` `InternPool.Index` which can then be used for the
generic function call.
Some other nice things made by this simplification are the removal of
`comptime_args_fn_inst` and `preallocated_new_func` from `Sema`, and the
messy logic associated with them.
I have not yet been able to measure the perf of this against master
branch. On one hand, it reduces memory usage and pointer chasing of the
most heavily used `InternPool` Tag - function bodies - but on the other
hand, it does evaluate function prototype expressions more than before.
We will soon find out.
2023-07-03 22:09:30 -07:00
|
|
|
|
2023-05-29 05:07:17 +01:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-05-22 07:58:02 -04:00
|
|
|
|
2023-05-29 05:07:17 +01:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
rework generic function calls
Abridged summary:
* Move `Module.Fn` into `InternPool`.
* Delete a lot of confusing and problematic `Sema` logic related to
generic function calls.
This commit removes `Module.Fn` and replaces it with two new
`InternPool.Tag` values:
* `func_decl` - corresponding to a function declared in the source
code. This one contains line/column numbers, zir_body_inst, etc.
* `func_instance` - one for each monomorphization of a generic
function. Contains a reference to the `func_decl` from whence the
instantiation came, along with the `comptime` parameter values (or
types in the case of `anytype`)
Since `InternPool` provides deduplication on these values, these fields
are now deleted from `Module`:
* `monomorphed_func_keys`
* `monomorphed_funcs`
* `align_stack_fns`
Instead of these, Sema logic for generic function instantiation now
unconditionally evaluates the function prototype expression for every
generic callsite. This is technically required in order for type
coercions to work. The previous code had some dubious, probably wrong
hacks to make things work, such as `hashUncoerced`. I'm not 100% sure
how we were able to eliminate that function and still pass all the
behavior tests, but I'm pretty sure things were still broken without
doing type coercion for every generic function call argument.
After the function prototype is evaluated, it produces a deduplicated
`func_instance` `InternPool.Index` which can then be used for the
generic function call.
Some other nice things made by this simplification are the removal of
`comptime_args_fn_inst` and `preallocated_new_func` from `Sema`, and the
messy logic associated with them.
I have not yet been able to measure the perf of this against master
branch. On one hand, it reduces memory usage and pointer chasing of the
most heavily used `InternPool` Tag - function bodies - but on the other
hand, it does evaluate function prototype expressions more than before.
We will soon find out.
2023-07-03 22:09:30 -07:00
|
|
|
|
2023-05-29 05:07:17 +01:00
|
|
|
|
|
|
|
|
|
2021-10-17 18:57:54 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2022-04-19 21:51:08 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-10-17 18:57:54 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-04-10 00:00:35 -04:00
|
|
|
|
2021-10-17 18:57:54 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2022-04-19 21:51:08 -07:00
|
|
|
|
2021-10-17 18:57:54 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-10-06 21:29:08 -07:00
|
|
|
|
|
|
|
|
|
2021-06-21 13:27:52 -07:00
|
|
|
|
2021-04-26 20:41:07 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-05-11 17:34:13 -07:00
|
|
|
|
|
|
|
|
|
2021-10-01 17:45:13 -05:00
|
|
|
|
2021-04-26 20:41:07 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2022-04-19 21:51:08 -07:00
|
|
|
|
2021-04-26 20:41:07 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-05-10 12:16:24 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-09-14 19:59:58 -07:00
|
|
|
|
2023-05-10 12:16:24 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2022-04-19 21:51:08 -07:00
|
|
|
|
2023-05-10 12:16:24 -07:00
|
|
|
|
compiler: eliminate Decl.value_arena and Sema.perm_arena
The main motivation for this commit is eliminating Decl.value_arena.
Everything else is dominoes.
Decl.name used to be stored in the GPA, now it is stored in InternPool.
It ended up being simpler to migrate other strings to be interned as
well, such as struct field names, union field names, and a few others.
This ended up requiring a big diff, sorry about that. But the changes
are pretty nice, we finally start to take advantage of InternPool's
existence.
global_error_set and error_name_list are simplified. Now it is a single
ArrayHashMap(NullTerminatedString, void) and the index is the error tag
value.
Module.tmp_hack_arena is re-introduced (it was removed in
eeff407941560ce8eb5b737b2436dfa93cfd3a0c) in order to deal with
comptime_args, optimized_order, and struct and union fields. After
structs and unions get moved into InternPool properly, tmp_hack_arena
can be deleted again.
2023-06-01 23:46:04 -07:00
|
|
|
|
2023-08-24 20:43:43 -07:00
|
|
|
|
2021-05-01 21:57:52 -07:00
|
|
|
|
2021-04-30 11:07:31 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-09-02 14:46:31 +02:00
|
|
|
|
2023-05-10 12:16:24 -07:00
|
|
|
|
2023-06-20 14:02:04 -04:00
|
|
|
|
compiler: eliminate Decl.value_arena and Sema.perm_arena
The main motivation for this commit is eliminating Decl.value_arena.
Everything else is dominoes.
Decl.name used to be stored in the GPA, now it is stored in InternPool.
It ended up being simpler to migrate other strings to be interned as
well, such as struct field names, union field names, and a few others.
This ended up requiring a big diff, sorry about that. But the changes
are pretty nice, we finally start to take advantage of InternPool's
existence.
global_error_set and error_name_list are simplified. Now it is a single
ArrayHashMap(NullTerminatedString, void) and the index is the error tag
value.
Module.tmp_hack_arena is re-introduced (it was removed in
eeff407941560ce8eb5b737b2436dfa93cfd3a0c) in order to deal with
comptime_args, optimized_order, and struct and union fields. After
structs and unions get moved into InternPool properly, tmp_hack_arena
can be deleted again.
2023-06-01 23:46:04 -07:00
|
|
|
|
2021-04-30 11:07:31 -07:00
|
|
|
|
2021-05-11 14:17:52 -07:00
|
|
|
|
stage2: garbage collect unused anon decls
After this change, the frontend and backend cooperate to keep track of
which Decls are actually emitted into the machine code. When any backend
sees a `decl_ref` Value, it must mark the corresponding Decl `alive`
field to true.
This prevents unused comptime data from spilling into the output object
files. For example, if you do an `inline for` loop, previously, any
intermediate value calculations would have gone into the object file.
Now they are garbage collected immediately after the owner Decl has its
machine code generated.
In the frontend, when it is time to send a Decl to the linker, if it has
not been marked "alive" then it is deleted instead.
Additional improvements:
* Resolve type ABI layouts after successful semantic analysis of a
Decl. This is needed so that the backend has access to struct fields.
* Sema: fix incorrect logic in resolveMaybeUndefVal. It should return
"not comptime known" instead of a compile error for global variables.
* `Value.pointerDeref` now returns `null` in the case that the pointer
deref cannot happen at compile-time. This is true for global
variables, for example. Another example is if a comptime known
pointer has a hard coded address value.
* Binary arithmetic sets the requireRuntimeBlock source location to the
lhs_src or rhs_src as appropriate instead of on the operator node.
* Fix LLVM codegen for slice_elem_val which had the wrong logic for
when the operand was not a pointer.
As noted in the comment in the implementation of deleteUnusedDecl, a
future improvement will be to rework the frontend/linker interface to
remove the frontend's responsibility of calling allocateDeclIndexes.
I discovered some issues with the plan9 linker backend that are related
to this, and worked around them for now.
2021-07-29 19:30:37 -07:00
|
|
|
|
2021-05-11 17:34:13 -07:00
|
|
|
|
2021-04-30 11:07:31 -07:00
|
|
|
|
2021-04-27 18:36:12 -07:00
|
|
|
|
2023-08-24 20:43:43 -07:00
|
|
|
|
2021-05-11 17:34:13 -07:00
|
|
|
|
2023-08-24 20:43:43 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-10-06 21:29:08 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-08-24 20:43:43 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-05-11 17:34:13 -07:00
|
|
|
|
2021-04-26 20:41:07 -07:00
|
|
|
|
|
|
|
|
|
2021-02-11 23:29:55 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2022-04-19 21:51:08 -07:00
|
|
|
|
2020-09-13 19:17:58 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2022-04-19 21:51:08 -07:00
|
|
|
|
|
|
|
|
|
2023-05-08 16:52:59 -07:00
|
|
|
|
2021-05-11 17:34:13 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-04-27 18:36:12 -07:00
|
|
|
|
2023-05-08 16:52:59 -07:00
|
|
|
|
2021-05-06 17:20:45 -07:00
|
|
|
|
2021-04-27 18:36:12 -07:00
|
|
|
|
2023-08-14 13:06:47 +01:00
|
|
|
|
|
|
|
|
|
2023-10-06 21:29:08 -07:00
|
|
|
|
|
|
|
|
|
2023-08-14 13:06:47 +01:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-04-27 18:36:12 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-06-02 01:55:16 +01:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-04-27 18:36:12 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
compiler: eliminate Decl.value_arena and Sema.perm_arena
The main motivation for this commit is eliminating Decl.value_arena.
Everything else is dominoes.
Decl.name used to be stored in the GPA, now it is stored in InternPool.
It ended up being simpler to migrate other strings to be interned as
well, such as struct field names, union field names, and a few others.
This ended up requiring a big diff, sorry about that. But the changes
are pretty nice, we finally start to take advantage of InternPool's
existence.
global_error_set and error_name_list are simplified. Now it is a single
ArrayHashMap(NullTerminatedString, void) and the index is the error tag
value.
Module.tmp_hack_arena is re-introduced (it was removed in
eeff407941560ce8eb5b737b2436dfa93cfd3a0c) in order to deal with
comptime_args, optimized_order, and struct and union fields. After
structs and unions get moved into InternPool properly, tmp_hack_arena
can be deleted again.
2023-06-01 23:46:04 -07:00
|
|
|
|
2021-04-27 18:36:12 -07:00
|
|
|
|
|
|
|
|
|
2022-04-19 21:51:08 -07:00
|
|
|
|
2023-05-22 07:58:02 -04:00
|
|
|
|
2023-07-31 01:50:54 -04:00
|
|
|
|
2022-02-09 00:10:53 -07:00
|
|
|
|
2023-07-08 23:39:37 -07:00
|
|
|
|
2023-05-22 07:58:02 -04:00
|
|
|
|
2023-06-02 01:55:16 +01:00
|
|
|
|
2023-08-14 13:06:47 +01:00
|
|
|
|
2021-04-27 18:36:12 -07:00
|
|
|
|
2021-05-17 17:39:52 -07:00
|
|
|
|
2021-05-06 17:20:45 -07:00
|
|
|
|
2023-09-19 23:50:04 -07:00
|
|
|
|
2021-05-06 17:20:45 -07:00
|
|
|
|
2021-10-01 17:45:13 -05:00
|
|
|
|
2021-04-27 18:36:12 -07:00
|
|
|
|
|
|
|
|
|
2022-04-19 21:51:08 -07:00
|
|
|
|
2021-10-01 12:05:12 -05:00
|
|
|
|
2023-09-14 19:59:58 -07:00
|
|
|
|
2021-04-27 18:36:12 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-07-09 15:31:47 -07:00
|
|
|
|
2021-04-27 18:36:12 -07:00
|
|
|
|
2023-05-08 16:52:59 -07:00
|
|
|
|
2021-04-28 22:43:26 -07:00
|
|
|
|
2021-04-27 18:36:12 -07:00
|
|
|
|
|
|
|
|
|
2022-03-09 12:24:18 +02:00
|
|
|
|
2023-08-14 13:06:47 +01:00
|
|
|
|
|
|
|
|
|
2023-06-02 01:55:16 +01:00
|
|
|
|
|
|
|
|
|
2023-07-10 10:52:17 -04:00
|
|
|
|
2023-06-02 01:55:16 +01:00
|
|
|
|
2022-07-28 17:28:37 +03:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-08-26 16:55:32 -04:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-09-01 17:33:45 +02:00
|
|
|
|
2021-08-20 15:23:55 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2022-11-12 18:23:40 +02:00
|
|
|
|
2021-04-27 18:36:12 -07:00
|
|
|
|
2023-01-10 21:54:58 +02:00
|
|
|
|
2022-04-19 21:51:08 -07:00
|
|
|
|
2022-07-28 17:28:37 +03:00
|
|
|
|
2022-04-19 21:51:08 -07:00
|
|
|
|
2022-03-22 00:23:54 -07:00
|
|
|
|
2021-08-28 15:35:59 -07:00
|
|
|
|
2023-05-20 12:09:07 -07:00
|
|
|
|
2023-05-08 16:52:59 -07:00
|
|
|
|
2022-07-28 17:28:37 +03:00
|
|
|
|
2021-08-28 15:35:59 -07:00
|
|
|
|
|
|
|
|
|
2023-05-21 04:29:34 -04:00
|
|
|
|
|
|
|
|
|
2023-06-20 14:02:04 -04:00
|
|
|
|
compiler: eliminate Decl.value_arena and Sema.perm_arena
The main motivation for this commit is eliminating Decl.value_arena.
Everything else is dominoes.
Decl.name used to be stored in the GPA, now it is stored in InternPool.
It ended up being simpler to migrate other strings to be interned as
well, such as struct field names, union field names, and a few others.
This ended up requiring a big diff, sorry about that. But the changes
are pretty nice, we finally start to take advantage of InternPool's
existence.
global_error_set and error_name_list are simplified. Now it is a single
ArrayHashMap(NullTerminatedString, void) and the index is the error tag
value.
Module.tmp_hack_arena is re-introduced (it was removed in
eeff407941560ce8eb5b737b2436dfa93cfd3a0c) in order to deal with
comptime_args, optimized_order, and struct and union fields. After
structs and unions get moved into InternPool properly, tmp_hack_arena
can be deleted again.
2023-06-01 23:46:04 -07:00
|
|
|
|
2021-08-28 15:35:59 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
rework generic function calls
Abridged summary:
* Move `Module.Fn` into `InternPool`.
* Delete a lot of confusing and problematic `Sema` logic related to
generic function calls.
This commit removes `Module.Fn` and replaces it with two new
`InternPool.Tag` values:
* `func_decl` - corresponding to a function declared in the source
code. This one contains line/column numbers, zir_body_inst, etc.
* `func_instance` - one for each monomorphization of a generic
function. Contains a reference to the `func_decl` from whence the
instantiation came, along with the `comptime` parameter values (or
types in the case of `anytype`)
Since `InternPool` provides deduplication on these values, these fields
are now deleted from `Module`:
* `monomorphed_func_keys`
* `monomorphed_funcs`
* `align_stack_fns`
Instead of these, Sema logic for generic function instantiation now
unconditionally evaluates the function prototype expression for every
generic callsite. This is technically required in order for type
coercions to work. The previous code had some dubious, probably wrong
hacks to make things work, such as `hashUncoerced`. I'm not 100% sure
how we were able to eliminate that function and still pass all the
behavior tests, but I'm pretty sure things were still broken without
doing type coercion for every generic function call argument.
After the function prototype is evaluated, it produces a deduplicated
`func_instance` `InternPool.Index` which can then be used for the
generic function call.
Some other nice things made by this simplification are the removal of
`comptime_args_fn_inst` and `preallocated_new_func` from `Sema`, and the
messy logic associated with them.
I have not yet been able to measure the perf of this against master
branch. On one hand, it reduces memory usage and pointer chasing of the
most heavily used `InternPool` Tag - function bodies - but on the other
hand, it does evaluate function prototype expressions more than before.
We will soon find out.
2023-07-03 22:09:30 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-04-28 22:43:26 -07:00
|
|
|
|
rework generic function calls
Abridged summary:
* Move `Module.Fn` into `InternPool`.
* Delete a lot of confusing and problematic `Sema` logic related to
generic function calls.
This commit removes `Module.Fn` and replaces it with two new
`InternPool.Tag` values:
* `func_decl` - corresponding to a function declared in the source
code. This one contains line/column numbers, zir_body_inst, etc.
* `func_instance` - one for each monomorphization of a generic
function. Contains a reference to the `func_decl` from whence the
instantiation came, along with the `comptime` parameter values (or
types in the case of `anytype`)
Since `InternPool` provides deduplication on these values, these fields
are now deleted from `Module`:
* `monomorphed_func_keys`
* `monomorphed_funcs`
* `align_stack_fns`
Instead of these, Sema logic for generic function instantiation now
unconditionally evaluates the function prototype expression for every
generic callsite. This is technically required in order for type
coercions to work. The previous code had some dubious, probably wrong
hacks to make things work, such as `hashUncoerced`. I'm not 100% sure
how we were able to eliminate that function and still pass all the
behavior tests, but I'm pretty sure things were still broken without
doing type coercion for every generic function call argument.
After the function prototype is evaluated, it produces a deduplicated
`func_instance` `InternPool.Index` which can then be used for the
generic function call.
Some other nice things made by this simplification are the removal of
`comptime_args_fn_inst` and `preallocated_new_func` from `Sema`, and the
messy logic associated with them.
I have not yet been able to measure the perf of this against master
branch. On one hand, it reduces memory usage and pointer chasing of the
most heavily used `InternPool` Tag - function bodies - but on the other
hand, it does evaluate function prototype expressions more than before.
We will soon find out.
2023-07-03 22:09:30 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-07-20 15:22:37 -07:00
|
|
|
|
2021-04-28 22:43:26 -07:00
|
|
|
|
rework generic function calls
Abridged summary:
* Move `Module.Fn` into `InternPool`.
* Delete a lot of confusing and problematic `Sema` logic related to
generic function calls.
This commit removes `Module.Fn` and replaces it with two new
`InternPool.Tag` values:
* `func_decl` - corresponding to a function declared in the source
code. This one contains line/column numbers, zir_body_inst, etc.
* `func_instance` - one for each monomorphization of a generic
function. Contains a reference to the `func_decl` from whence the
instantiation came, along with the `comptime` parameter values (or
types in the case of `anytype`)
Since `InternPool` provides deduplication on these values, these fields
are now deleted from `Module`:
* `monomorphed_func_keys`
* `monomorphed_funcs`
* `align_stack_fns`
Instead of these, Sema logic for generic function instantiation now
unconditionally evaluates the function prototype expression for every
generic callsite. This is technically required in order for type
coercions to work. The previous code had some dubious, probably wrong
hacks to make things work, such as `hashUncoerced`. I'm not 100% sure
how we were able to eliminate that function and still pass all the
behavior tests, but I'm pretty sure things were still broken without
doing type coercion for every generic function call argument.
After the function prototype is evaluated, it produces a deduplicated
`func_instance` `InternPool.Index` which can then be used for the
generic function call.
Some other nice things made by this simplification are the removal of
`comptime_args_fn_inst` and `preallocated_new_func` from `Sema`, and the
messy logic associated with them.
I have not yet been able to measure the perf of this against master
branch. On one hand, it reduces memory usage and pointer chasing of the
most heavily used `InternPool` Tag - function bodies - but on the other
hand, it does evaluate function prototype expressions more than before.
We will soon find out.
2023-07-03 22:09:30 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-04-28 22:43:26 -07:00
|
|
|
|
rework generic function calls
Abridged summary:
* Move `Module.Fn` into `InternPool`.
* Delete a lot of confusing and problematic `Sema` logic related to
generic function calls.
This commit removes `Module.Fn` and replaces it with two new
`InternPool.Tag` values:
* `func_decl` - corresponding to a function declared in the source
code. This one contains line/column numbers, zir_body_inst, etc.
* `func_instance` - one for each monomorphization of a generic
function. Contains a reference to the `func_decl` from whence the
instantiation came, along with the `comptime` parameter values (or
types in the case of `anytype`)
Since `InternPool` provides deduplication on these values, these fields
are now deleted from `Module`:
* `monomorphed_func_keys`
* `monomorphed_funcs`
* `align_stack_fns`
Instead of these, Sema logic for generic function instantiation now
unconditionally evaluates the function prototype expression for every
generic callsite. This is technically required in order for type
coercions to work. The previous code had some dubious, probably wrong
hacks to make things work, such as `hashUncoerced`. I'm not 100% sure
how we were able to eliminate that function and still pass all the
behavior tests, but I'm pretty sure things were still broken without
doing type coercion for every generic function call argument.
After the function prototype is evaluated, it produces a deduplicated
`func_instance` `InternPool.Index` which can then be used for the
generic function call.
Some other nice things made by this simplification are the removal of
`comptime_args_fn_inst` and `preallocated_new_func` from `Sema`, and the
messy logic associated with them.
I have not yet been able to measure the perf of this against master
branch. On one hand, it reduces memory usage and pointer chasing of the
most heavily used `InternPool` Tag - function bodies - but on the other
hand, it does evaluate function prototype expressions more than before.
We will soon find out.
2023-07-03 22:09:30 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-07-20 15:22:37 -07:00
|
|
|
|
rework generic function calls
Abridged summary:
* Move `Module.Fn` into `InternPool`.
* Delete a lot of confusing and problematic `Sema` logic related to
generic function calls.
This commit removes `Module.Fn` and replaces it with two new
`InternPool.Tag` values:
* `func_decl` - corresponding to a function declared in the source
code. This one contains line/column numbers, zir_body_inst, etc.
* `func_instance` - one for each monomorphization of a generic
function. Contains a reference to the `func_decl` from whence the
instantiation came, along with the `comptime` parameter values (or
types in the case of `anytype`)
Since `InternPool` provides deduplication on these values, these fields
are now deleted from `Module`:
* `monomorphed_func_keys`
* `monomorphed_funcs`
* `align_stack_fns`
Instead of these, Sema logic for generic function instantiation now
unconditionally evaluates the function prototype expression for every
generic callsite. This is technically required in order for type
coercions to work. The previous code had some dubious, probably wrong
hacks to make things work, such as `hashUncoerced`. I'm not 100% sure
how we were able to eliminate that function and still pass all the
behavior tests, but I'm pretty sure things were still broken without
doing type coercion for every generic function call argument.
After the function prototype is evaluated, it produces a deduplicated
`func_instance` `InternPool.Index` which can then be used for the
generic function call.
Some other nice things made by this simplification are the removal of
`comptime_args_fn_inst` and `preallocated_new_func` from `Sema`, and the
messy logic associated with them.
I have not yet been able to measure the perf of this against master
branch. On one hand, it reduces memory usage and pointer chasing of the
most heavily used `InternPool` Tag - function bodies - but on the other
hand, it does evaluate function prototype expressions more than before.
We will soon find out.
2023-07-03 22:09:30 -07:00
|
|
|
|
2021-04-28 22:43:26 -07:00
|
|
|
|
rework generic function calls
Abridged summary:
* Move `Module.Fn` into `InternPool`.
* Delete a lot of confusing and problematic `Sema` logic related to
generic function calls.
This commit removes `Module.Fn` and replaces it with two new
`InternPool.Tag` values:
* `func_decl` - corresponding to a function declared in the source
code. This one contains line/column numbers, zir_body_inst, etc.
* `func_instance` - one for each monomorphization of a generic
function. Contains a reference to the `func_decl` from whence the
instantiation came, along with the `comptime` parameter values (or
types in the case of `anytype`)
Since `InternPool` provides deduplication on these values, these fields
are now deleted from `Module`:
* `monomorphed_func_keys`
* `monomorphed_funcs`
* `align_stack_fns`
Instead of these, Sema logic for generic function instantiation now
unconditionally evaluates the function prototype expression for every
generic callsite. This is technically required in order for type
coercions to work. The previous code had some dubious, probably wrong
hacks to make things work, such as `hashUncoerced`. I'm not 100% sure
how we were able to eliminate that function and still pass all the
behavior tests, but I'm pretty sure things were still broken without
doing type coercion for every generic function call argument.
After the function prototype is evaluated, it produces a deduplicated
`func_instance` `InternPool.Index` which can then be used for the
generic function call.
Some other nice things made by this simplification are the removal of
`comptime_args_fn_inst` and `preallocated_new_func` from `Sema`, and the
messy logic associated with them.
I have not yet been able to measure the perf of this against master
branch. On one hand, it reduces memory usage and pointer chasing of the
most heavily used `InternPool` Tag - function bodies - but on the other
hand, it does evaluate function prototype expressions more than before.
We will soon find out.
2023-07-03 22:09:30 -07:00
|
|
|
|
|
|
|
|
|
2021-04-27 18:36:12 -07:00
|
|
|
|
2021-07-20 15:22:37 -07:00
|
|
|
|
|
|
|
|
|
2022-04-19 21:51:08 -07:00
|
|
|
|
2021-07-20 15:22:37 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2022-01-24 20:38:56 -07:00
|
|
|
|
2023-05-25 05:47:25 -04:00
|
|
|
|
2021-11-24 22:27:40 -07:00
|
|
|
|
2023-05-04 21:40:35 -07:00
|
|
|
|
rework generic function calls
Abridged summary:
* Move `Module.Fn` into `InternPool`.
* Delete a lot of confusing and problematic `Sema` logic related to
generic function calls.
This commit removes `Module.Fn` and replaces it with two new
`InternPool.Tag` values:
* `func_decl` - corresponding to a function declared in the source
code. This one contains line/column numbers, zir_body_inst, etc.
* `func_instance` - one for each monomorphization of a generic
function. Contains a reference to the `func_decl` from whence the
instantiation came, along with the `comptime` parameter values (or
types in the case of `anytype`)
Since `InternPool` provides deduplication on these values, these fields
are now deleted from `Module`:
* `monomorphed_func_keys`
* `monomorphed_funcs`
* `align_stack_fns`
Instead of these, Sema logic for generic function instantiation now
unconditionally evaluates the function prototype expression for every
generic callsite. This is technically required in order for type
coercions to work. The previous code had some dubious, probably wrong
hacks to make things work, such as `hashUncoerced`. I'm not 100% sure
how we were able to eliminate that function and still pass all the
behavior tests, but I'm pretty sure things were still broken without
doing type coercion for every generic function call argument.
After the function prototype is evaluated, it produces a deduplicated
`func_instance` `InternPool.Index` which can then be used for the
generic function call.
Some other nice things made by this simplification are the removal of
`comptime_args_fn_inst` and `preallocated_new_func` from `Sema`, and the
messy logic associated with them.
I have not yet been able to measure the perf of this against master
branch. On one hand, it reduces memory usage and pointer chasing of the
most heavily used `InternPool` Tag - function bodies - but on the other
hand, it does evaluate function prototype expressions more than before.
We will soon find out.
2023-07-03 22:09:30 -07:00
|
|
|
|
2023-05-22 07:58:02 -04:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-05-03 19:12:53 -07:00
|
|
|
|
2021-11-24 22:27:40 -07:00
|
|
|
|
2023-05-22 07:58:02 -04:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-05-03 19:12:53 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-07-20 15:22:37 -07:00
|
|
|
|
|
|
|
|
|
2023-05-20 12:09:07 -07:00
|
|
|
|
compiler: eliminate Decl.value_arena and Sema.perm_arena
The main motivation for this commit is eliminating Decl.value_arena.
Everything else is dominoes.
Decl.name used to be stored in the GPA, now it is stored in InternPool.
It ended up being simpler to migrate other strings to be interned as
well, such as struct field names, union field names, and a few others.
This ended up requiring a big diff, sorry about that. But the changes
are pretty nice, we finally start to take advantage of InternPool's
existence.
global_error_set and error_name_list are simplified. Now it is a single
ArrayHashMap(NullTerminatedString, void) and the index is the error tag
value.
Module.tmp_hack_arena is re-introduced (it was removed in
eeff407941560ce8eb5b737b2436dfa93cfd3a0c) in order to deal with
comptime_args, optimized_order, and struct and union fields. After
structs and unions get moved into InternPool properly, tmp_hack_arena
can be deleted again.
2023-06-01 23:46:04 -07:00
|
|
|
|
2023-06-20 14:02:04 -04:00
|
|
|
|
2023-05-08 16:52:59 -07:00
|
|
|
|
2023-06-20 14:02:04 -04:00
|
|
|
|
2022-10-18 07:02:10 -04:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-05-08 16:52:59 -07:00
|
|
|
|
compiler: eliminate Decl.value_arena and Sema.perm_arena
The main motivation for this commit is eliminating Decl.value_arena.
Everything else is dominoes.
Decl.name used to be stored in the GPA, now it is stored in InternPool.
It ended up being simpler to migrate other strings to be interned as
well, such as struct field names, union field names, and a few others.
This ended up requiring a big diff, sorry about that. But the changes
are pretty nice, we finally start to take advantage of InternPool's
existence.
global_error_set and error_name_list are simplified. Now it is a single
ArrayHashMap(NullTerminatedString, void) and the index is the error tag
value.
Module.tmp_hack_arena is re-introduced (it was removed in
eeff407941560ce8eb5b737b2436dfa93cfd3a0c) in order to deal with
comptime_args, optimized_order, and struct and union fields. After
structs and unions get moved into InternPool properly, tmp_hack_arena
can be deleted again.
2023-06-01 23:46:04 -07:00
|
|
|
|
2023-08-26 16:55:32 -04:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2022-10-18 07:02:10 -04:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
rework generic function calls
Abridged summary:
* Move `Module.Fn` into `InternPool`.
* Delete a lot of confusing and problematic `Sema` logic related to
generic function calls.
This commit removes `Module.Fn` and replaces it with two new
`InternPool.Tag` values:
* `func_decl` - corresponding to a function declared in the source
code. This one contains line/column numbers, zir_body_inst, etc.
* `func_instance` - one for each monomorphization of a generic
function. Contains a reference to the `func_decl` from whence the
instantiation came, along with the `comptime` parameter values (or
types in the case of `anytype`)
Since `InternPool` provides deduplication on these values, these fields
are now deleted from `Module`:
* `monomorphed_func_keys`
* `monomorphed_funcs`
* `align_stack_fns`
Instead of these, Sema logic for generic function instantiation now
unconditionally evaluates the function prototype expression for every
generic callsite. This is technically required in order for type
coercions to work. The previous code had some dubious, probably wrong
hacks to make things work, such as `hashUncoerced`. I'm not 100% sure
how we were able to eliminate that function and still pass all the
behavior tests, but I'm pretty sure things were still broken without
doing type coercion for every generic function call argument.
After the function prototype is evaluated, it produces a deduplicated
`func_instance` `InternPool.Index` which can then be used for the
generic function call.
Some other nice things made by this simplification are the removal of
`comptime_args_fn_inst` and `preallocated_new_func` from `Sema`, and the
messy logic associated with them.
I have not yet been able to measure the perf of this against master
branch. On one hand, it reduces memory usage and pointer chasing of the
most heavily used `InternPool` Tag - function bodies - but on the other
hand, it does evaluate function prototype expressions more than before.
We will soon find out.
2023-07-03 22:09:30 -07:00
|
|
|
|
compiler: eliminate Decl.value_arena and Sema.perm_arena
The main motivation for this commit is eliminating Decl.value_arena.
Everything else is dominoes.
Decl.name used to be stored in the GPA, now it is stored in InternPool.
It ended up being simpler to migrate other strings to be interned as
well, such as struct field names, union field names, and a few others.
This ended up requiring a big diff, sorry about that. But the changes
are pretty nice, we finally start to take advantage of InternPool's
existence.
global_error_set and error_name_list are simplified. Now it is a single
ArrayHashMap(NullTerminatedString, void) and the index is the error tag
value.
Module.tmp_hack_arena is re-introduced (it was removed in
eeff407941560ce8eb5b737b2436dfa93cfd3a0c) in order to deal with
comptime_args, optimized_order, and struct and union fields. After
structs and unions get moved into InternPool properly, tmp_hack_arena
can be deleted again.
2023-06-01 23:46:04 -07:00
|
|
|
|
2022-10-18 07:02:10 -04:00
|
|
|
|
|
|
|
|
|
rework generic function calls
Abridged summary:
* Move `Module.Fn` into `InternPool`.
* Delete a lot of confusing and problematic `Sema` logic related to
generic function calls.
This commit removes `Module.Fn` and replaces it with two new
`InternPool.Tag` values:
* `func_decl` - corresponding to a function declared in the source
code. This one contains line/column numbers, zir_body_inst, etc.
* `func_instance` - one for each monomorphization of a generic
function. Contains a reference to the `func_decl` from whence the
instantiation came, along with the `comptime` parameter values (or
types in the case of `anytype`)
Since `InternPool` provides deduplication on these values, these fields
are now deleted from `Module`:
* `monomorphed_func_keys`
* `monomorphed_funcs`
* `align_stack_fns`
Instead of these, Sema logic for generic function instantiation now
unconditionally evaluates the function prototype expression for every
generic callsite. This is technically required in order for type
coercions to work. The previous code had some dubious, probably wrong
hacks to make things work, such as `hashUncoerced`. I'm not 100% sure
how we were able to eliminate that function and still pass all the
behavior tests, but I'm pretty sure things were still broken without
doing type coercion for every generic function call argument.
After the function prototype is evaluated, it produces a deduplicated
`func_instance` `InternPool.Index` which can then be used for the
generic function call.
Some other nice things made by this simplification are the removal of
`comptime_args_fn_inst` and `preallocated_new_func` from `Sema`, and the
messy logic associated with them.
I have not yet been able to measure the perf of this against master
branch. On one hand, it reduces memory usage and pointer chasing of the
most heavily used `InternPool` Tag - function bodies - but on the other
hand, it does evaluate function prototype expressions more than before.
We will soon find out.
2023-07-03 22:09:30 -07:00
|
|
|
|
2023-05-22 07:58:02 -04:00
|
|
|
|
|
|
|
|
|
2022-10-18 07:02:10 -04:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-05-08 16:52:59 -07:00
|
|
|
|
2022-10-18 07:02:10 -04:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-07-20 15:22:37 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2022-01-24 20:38:56 -07:00
|
|
|
|
2022-11-12 18:23:40 +02:00
|
|
|
|
2022-01-24 20:38:56 -07:00
|
|
|
|
|
|
|
|
|
stage2: implement runtime pointer access to global constants
The main problem that motivated these changes is that global constants
which are referenced by pointer would not be emitted into the binary.
This happened because `semaDecl` did not add `codegen_decl` tasks for
global constants, instead relying on the constant values being copied as
necessary. However when the global constants are referenced by pointer,
they need to be sent to the linker to be emitted.
After making global const arrays, structs, and unions get emitted, this
uncovered a latent issue: the anonymous decls that they referenced would
get garbage collected (via `deleteUnusedDecl`) even though they would
later be referenced by the global const.
In order to solve this problem, I introduced `anon_work_queue` which is
the same as `work_queue` except a lower priority. The `codegen_decl`
task for anon decls goes into the `anon_work_queue` ensuring that the
owner decl gets a chance to mark its anon decls as alive before they are
possibly deleted.
This caused a few regressions, which I made the judgement call to add
workarounds for. Two steps forward, one step back, is still progress.
The regressions were:
* Two behavior tests having to do with unions. These tests were
intentionally exercising the LLVM constant value lowering, however,
due to the bug with garbage collection that was fixed in this commit,
the LLVM code was not getting exercised, and union types/values were
not implemented correctly, due to me forgetting that LLVM does not
allow bitcasting aggregate values.
- This is worked around by allowing those 2 test cases to regress,
moving them to the "passing for stage1 only" section.
* The test-stage2 test cases (in test/cases/*) for non-LLVM backends
previously did not have any calls to lower struct values, but now
they do. The code that was there was just `@panic("TODO")`. I
replaced that code with a stub that generates the wrong value. This
is an intentional miscompilation that will obviously need to get
fixed before any struct behavior tests pass. None of the current
tests we have exercise loading any values from these global const
structs, so there is not a problem until we try to improve these
backends.
2021-10-26 22:29:43 -07:00
|
|
|
|
2022-03-26 00:33:14 -07:00
|
|
|
|
|
|
|
|
|
2022-11-12 18:23:40 +02:00
|
|
|
|
2022-03-26 00:33:14 -07:00
|
|
|
|
2022-04-19 21:51:08 -07:00
|
|
|
|
2021-07-20 15:22:37 -07:00
|
|
|
|
|
|
|
|
|
2022-04-19 21:51:08 -07:00
|
|
|
|
2021-07-20 15:22:37 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-06-15 13:14:16 +06:00
|
|
|
|
2021-07-20 15:22:37 -07:00
|
|
|
|
2023-06-12 01:44:12 -04:00
|
|
|
|
2021-07-20 15:22:37 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-02-11 23:29:55 -07:00
|
|
|
|
|
|
|
|
|
stage2: implement simple enums
A simple enum is an enum which has an automatic integer tag type,
all tag values automatically assigned, and no top level declarations.
Such enums are created directly in AstGen and shared by all the
generic/comptime instantiations of the surrounding ZIR code. This
commit implements, but does not yet add any test cases for, simple enums.
A full enum is an enum for which any of the above conditions are not
true. Full enums are created in Sema, and therefore will create a unique
type per generic/comptime instantiation. This commit does not implement
full enums. However the `enum_decl_nonexhaustive` ZIR instruction is
added and the respective Type functions are filled out.
This commit makes an improvement to ZIR code, removing the decls array
and removing the decl_map from AstGen. Instead, decl_ref and
decl_val ZIR instructions index into the `owner_decl.dependencies`
ArrayHashMap. We already need this dependencies array for incremental
compilation purposes, and so repurposing it to also use it for ZIR decl
indexes makes for efficient memory usage.
Similarly, this commit fixes up incorrect memory management by removing
the `const` ZIR instruction. The two places it was used stored memory in
the AstGen arena, which may get freed after Sema. Now it properly sets
up a new anonymous Decl for error sets and uses a normal decl_val
instruction.
The other usage of `const` ZIR instruction was float literals. These are
now changed to use `float` ZIR instruction when the value fits inside
`zir.Inst.Data` and `float128` otherwise.
AstGen + Sema: implement int_to_enum and enum_to_int. No tests yet; I expect to
have to make some fixes before they will pass tests. Will do that in the
branch before merging.
AstGen: fix struct astgen incorrectly counting decls as fields.
Type/Value: give up on trying to exhaustively list every tag all the
time. This makes the file more manageable. Also found a bug with
i128/u128 this way, since the name of the function was more obvious when
looking at the tag values.
Type: implement abiAlignment and abiSize for structs. This will need to
get more sophisticated at some point, but for now it is progress.
Value: add new `enum_field_index` tag.
Value: add hash_u32, needed when using ArrayHashMap.
2021-04-06 17:43:56 -07:00
|
|
|
|
2022-04-19 21:51:08 -07:00
|
|
|
|
2023-04-19 17:40:29 +01:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2022-04-19 21:51:08 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-05-14 17:41:22 -07:00
|
|
|
|
2023-04-19 17:40:29 +01:00
|
|
|
|
2023-06-15 13:14:16 +06:00
|
|
|
|
2023-04-19 17:40:29 +01:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-04-07 19:38:00 -07:00
|
|
|
|
|
|
|
|
|
2022-04-19 21:51:08 -07:00
|
|
|
|
2021-04-07 19:38:00 -07:00
|
|
|
|
|
|
|
|
|
2023-04-19 17:40:29 +01:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2020-09-13 19:17:58 -07:00
|
|
|
|
|
|
|
|
|
2021-04-16 17:28:28 -07:00
|
|
|
|
2021-10-01 17:45:13 -05:00
|
|
|
|
2021-04-16 17:28:28 -07:00
|
|
|
|
2022-11-27 14:17:55 +00:00
|
|
|
|
2021-04-16 17:28:28 -07:00
|
|
|
|
|
|
|
|
|
2023-10-06 21:29:08 -07:00
|
|
|
|
|
|
|
|
|
stage2: entry point via std lib and proper updated file detection
Instead of Module setting up the root_scope with the root source file,
instead, Module relies on the package table graph being set up properly,
and inside `update()`, it does the equivalent of `_ = @import("std");`.
This, in term, imports start.zig, which has the logic to call main (or
not). `Module` no longer has `root_scope` - the root source file is no
longer special, it's just in the package table mapped to "root".
I also went ahead and implemented proper detection of updated files.
mtime, inode, size, and source hash are kept in `Scope.File`.
During an update, iterate over `import_table` and stat each file to find
out which ones are updated.
The source hash is redundant with the source hash used by the struct
decl that corresponds to the file, so it should be removed in a future
commit before merging the branch.
* AstGen: add "previously declared here" notes for variables shadowing
decls.
* Parse imports as structs. Module now calls `AstGen.structDeclInner`,
which is called by `AstGen.containerDecl`.
- `importFile` is a bit kludgy with how it handles the top level Decl
that kinda gets merged into the struct decl at the end of the
function. Be on the look out for bugs related to that as well as
possibly cleaner ways to implement this.
* Module: factor out lookupDeclName into lookupIdentifier and lookupNa
* Rename `Scope.Container` to `Scope.Namespace`.
* Delete some dead code.
This branch won't work until `usingnamespace` is implemented because it
relies on `@import("builtin").OutputMode` and `OutputMode` comes from a
`usingnamespace`.
2021-04-09 23:17:50 -07:00
|
|
|
|
|
|
|
|
|
2021-04-16 19:45:58 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-10-06 21:29:08 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-04-16 19:45:58 -07:00
|
|
|
|
stage2: entry point via std lib and proper updated file detection
Instead of Module setting up the root_scope with the root source file,
instead, Module relies on the package table graph being set up properly,
and inside `update()`, it does the equivalent of `_ = @import("std");`.
This, in term, imports start.zig, which has the logic to call main (or
not). `Module` no longer has `root_scope` - the root source file is no
longer special, it's just in the package table mapped to "root".
I also went ahead and implemented proper detection of updated files.
mtime, inode, size, and source hash are kept in `Scope.File`.
During an update, iterate over `import_table` and stat each file to find
out which ones are updated.
The source hash is redundant with the source hash used by the struct
decl that corresponds to the file, so it should be removed in a future
commit before merging the branch.
* AstGen: add "previously declared here" notes for variables shadowing
decls.
* Parse imports as structs. Module now calls `AstGen.structDeclInner`,
which is called by `AstGen.containerDecl`.
- `importFile` is a bit kludgy with how it handles the top level Decl
that kinda gets merged into the struct decl at the end of the
function. Be on the look out for bugs related to that as well as
possibly cleaner ways to implement this.
* Module: factor out lookupDeclName into lookupIdentifier and lookupNa
* Rename `Scope.Container` to `Scope.Namespace`.
* Delete some dead code.
This branch won't work until `usingnamespace` is implemented because it
relies on `@import("builtin").OutputMode` and `OutputMode` comes from a
`usingnamespace`.
2021-04-09 23:17:50 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-11-24 19:09:45 -07:00
|
|
|
|
2023-02-17 01:44:08 +00:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
stage2: entry point via std lib and proper updated file detection
Instead of Module setting up the root_scope with the root source file,
instead, Module relies on the package table graph being set up properly,
and inside `update()`, it does the equivalent of `_ = @import("std");`.
This, in term, imports start.zig, which has the logic to call main (or
not). `Module` no longer has `root_scope` - the root source file is no
longer special, it's just in the package table mapped to "root".
I also went ahead and implemented proper detection of updated files.
mtime, inode, size, and source hash are kept in `Scope.File`.
During an update, iterate over `import_table` and stat each file to find
out which ones are updated.
The source hash is redundant with the source hash used by the struct
decl that corresponds to the file, so it should be removed in a future
commit before merging the branch.
* AstGen: add "previously declared here" notes for variables shadowing
decls.
* Parse imports as structs. Module now calls `AstGen.structDeclInner`,
which is called by `AstGen.containerDecl`.
- `importFile` is a bit kludgy with how it handles the top level Decl
that kinda gets merged into the struct decl at the end of the
function. Be on the look out for bugs related to that as well as
possibly cleaner ways to implement this.
* Module: factor out lookupDeclName into lookupIdentifier and lookupNa
* Rename `Scope.Container` to `Scope.Namespace`.
* Delete some dead code.
This branch won't work until `usingnamespace` is implemented because it
relies on `@import("builtin").OutputMode` and `OutputMode` comes from a
`usingnamespace`.
2021-04-09 23:17:50 -07:00
|
|
|
|
2021-04-16 19:45:58 -07:00
|
|
|
|
|
|
|
|
|
stage2: entry point via std lib and proper updated file detection
Instead of Module setting up the root_scope with the root source file,
instead, Module relies on the package table graph being set up properly,
and inside `update()`, it does the equivalent of `_ = @import("std");`.
This, in term, imports start.zig, which has the logic to call main (or
not). `Module` no longer has `root_scope` - the root source file is no
longer special, it's just in the package table mapped to "root".
I also went ahead and implemented proper detection of updated files.
mtime, inode, size, and source hash are kept in `Scope.File`.
During an update, iterate over `import_table` and stat each file to find
out which ones are updated.
The source hash is redundant with the source hash used by the struct
decl that corresponds to the file, so it should be removed in a future
commit before merging the branch.
* AstGen: add "previously declared here" notes for variables shadowing
decls.
* Parse imports as structs. Module now calls `AstGen.structDeclInner`,
which is called by `AstGen.containerDecl`.
- `importFile` is a bit kludgy with how it handles the top level Decl
that kinda gets merged into the struct decl at the end of the
function. Be on the look out for bugs related to that as well as
possibly cleaner ways to implement this.
* Module: factor out lookupDeclName into lookupIdentifier and lookupNa
* Rename `Scope.Container` to `Scope.Namespace`.
* Delete some dead code.
This branch won't work until `usingnamespace` is implemented because it
relies on `@import("builtin").OutputMode` and `OutputMode` comes from a
`usingnamespace`.
2021-04-09 23:17:50 -07:00
|
|
|
|
2021-10-01 17:45:13 -05:00
|
|
|
|
2021-04-16 19:45:58 -07:00
|
|
|
|
|
|
|
|
|
2021-11-24 19:09:45 -07:00
|
|
|
|
2021-06-03 15:39:26 -05:00
|
|
|
|
stage2: entry point via std lib and proper updated file detection
Instead of Module setting up the root_scope with the root source file,
instead, Module relies on the package table graph being set up properly,
and inside `update()`, it does the equivalent of `_ = @import("std");`.
This, in term, imports start.zig, which has the logic to call main (or
not). `Module` no longer has `root_scope` - the root source file is no
longer special, it's just in the package table mapped to "root".
I also went ahead and implemented proper detection of updated files.
mtime, inode, size, and source hash are kept in `Scope.File`.
During an update, iterate over `import_table` and stat each file to find
out which ones are updated.
The source hash is redundant with the source hash used by the struct
decl that corresponds to the file, so it should be removed in a future
commit before merging the branch.
* AstGen: add "previously declared here" notes for variables shadowing
decls.
* Parse imports as structs. Module now calls `AstGen.structDeclInner`,
which is called by `AstGen.containerDecl`.
- `importFile` is a bit kludgy with how it handles the top level Decl
that kinda gets merged into the struct decl at the end of the
function. Be on the look out for bugs related to that as well as
possibly cleaner ways to implement this.
* Module: factor out lookupDeclName into lookupIdentifier and lookupNa
* Rename `Scope.Container` to `Scope.Namespace`.
* Delete some dead code.
This branch won't work until `usingnamespace` is implemented because it
relies on `@import("builtin").OutputMode` and `OutputMode` comes from a
`usingnamespace`.
2021-04-09 23:17:50 -07:00
|
|
|
|
2021-04-16 19:45:58 -07:00
|
|
|
|
stage2: entry point via std lib and proper updated file detection
Instead of Module setting up the root_scope with the root source file,
instead, Module relies on the package table graph being set up properly,
and inside `update()`, it does the equivalent of `_ = @import("std");`.
This, in term, imports start.zig, which has the logic to call main (or
not). `Module` no longer has `root_scope` - the root source file is no
longer special, it's just in the package table mapped to "root".
I also went ahead and implemented proper detection of updated files.
mtime, inode, size, and source hash are kept in `Scope.File`.
During an update, iterate over `import_table` and stat each file to find
out which ones are updated.
The source hash is redundant with the source hash used by the struct
decl that corresponds to the file, so it should be removed in a future
commit before merging the branch.
* AstGen: add "previously declared here" notes for variables shadowing
decls.
* Parse imports as structs. Module now calls `AstGen.structDeclInner`,
which is called by `AstGen.containerDecl`.
- `importFile` is a bit kludgy with how it handles the top level Decl
that kinda gets merged into the struct decl at the end of the
function. Be on the look out for bugs related to that as well as
possibly cleaner ways to implement this.
* Module: factor out lookupDeclName into lookupIdentifier and lookupNa
* Rename `Scope.Container` to `Scope.Namespace`.
* Delete some dead code.
This branch won't work until `usingnamespace` is implemented because it
relies on `@import("builtin").OutputMode` and `OutputMode` comes from a
`usingnamespace`.
2021-04-09 23:17:50 -07:00
|
|
|
|
|
|
|
|
|
2021-04-14 11:26:53 -07:00
|
|
|
|
|
|
|
|
|
2021-12-30 20:49:02 -07:00
|
|
|
|
stage2: entry point via std lib and proper updated file detection
Instead of Module setting up the root_scope with the root source file,
instead, Module relies on the package table graph being set up properly,
and inside `update()`, it does the equivalent of `_ = @import("std");`.
This, in term, imports start.zig, which has the logic to call main (or
not). `Module` no longer has `root_scope` - the root source file is no
longer special, it's just in the package table mapped to "root".
I also went ahead and implemented proper detection of updated files.
mtime, inode, size, and source hash are kept in `Scope.File`.
During an update, iterate over `import_table` and stat each file to find
out which ones are updated.
The source hash is redundant with the source hash used by the struct
decl that corresponds to the file, so it should be removed in a future
commit before merging the branch.
* AstGen: add "previously declared here" notes for variables shadowing
decls.
* Parse imports as structs. Module now calls `AstGen.structDeclInner`,
which is called by `AstGen.containerDecl`.
- `importFile` is a bit kludgy with how it handles the top level Decl
that kinda gets merged into the struct decl at the end of the
function. Be on the look out for bugs related to that as well as
possibly cleaner ways to implement this.
* Module: factor out lookupDeclName into lookupIdentifier and lookupNa
* Rename `Scope.Container` to `Scope.Namespace`.
* Delete some dead code.
This branch won't work until `usingnamespace` is implemented because it
relies on `@import("builtin").OutputMode` and `OutputMode` comes from a
`usingnamespace`.
2021-04-09 23:17:50 -07:00
|
|
|
|
2021-04-14 11:26:53 -07:00
|
|
|
|
stage2: entry point via std lib and proper updated file detection
Instead of Module setting up the root_scope with the root source file,
instead, Module relies on the package table graph being set up properly,
and inside `update()`, it does the equivalent of `_ = @import("std");`.
This, in term, imports start.zig, which has the logic to call main (or
not). `Module` no longer has `root_scope` - the root source file is no
longer special, it's just in the package table mapped to "root".
I also went ahead and implemented proper detection of updated files.
mtime, inode, size, and source hash are kept in `Scope.File`.
During an update, iterate over `import_table` and stat each file to find
out which ones are updated.
The source hash is redundant with the source hash used by the struct
decl that corresponds to the file, so it should be removed in a future
commit before merging the branch.
* AstGen: add "previously declared here" notes for variables shadowing
decls.
* Parse imports as structs. Module now calls `AstGen.structDeclInner`,
which is called by `AstGen.containerDecl`.
- `importFile` is a bit kludgy with how it handles the top level Decl
that kinda gets merged into the struct decl at the end of the
function. Be on the look out for bugs related to that as well as
possibly cleaner ways to implement this.
* Module: factor out lookupDeclName into lookupIdentifier and lookupNa
* Rename `Scope.Container` to `Scope.Namespace`.
* Delete some dead code.
This branch won't work until `usingnamespace` is implemented because it
relies on `@import("builtin").OutputMode` and `OutputMode` comes from a
`usingnamespace`.
2021-04-09 23:17:50 -07:00
|
|
|
|
2023-10-06 21:29:08 -07:00
|
|
|
|
2022-04-19 21:51:08 -07:00
|
|
|
|
2021-04-16 19:45:58 -07:00
|
|
|
|
2022-11-27 14:17:55 +00:00
|
|
|
|
2021-04-16 19:45:58 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2022-11-27 14:17:55 +00:00
|
|
|
|
2021-04-16 19:45:58 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-10-01 17:45:13 -05:00
|
|
|
|
2021-04-16 19:45:58 -07:00
|
|
|
|
|
|
|
|
|
2022-06-14 14:36:19 -07:00
|
|
|
|
2023-10-06 21:29:08 -07:00
|
|
|
|
2022-06-14 14:36:19 -07:00
|
|
|
|
|
|
|
|
|
2023-10-06 21:29:08 -07:00
|
|
|
|
2022-06-14 14:36:19 -07:00
|
|
|
|
|
|
|
|
|
2023-10-06 17:05:58 -07:00
|
|
|
|
2022-06-14 14:36:19 -07:00
|
|
|
|
2023-10-06 21:29:08 -07:00
|
|
|
|
2021-06-21 13:27:52 -07:00
|
|
|
|
2021-04-16 19:45:58 -07:00
|
|
|
|
2021-06-22 16:11:02 -07:00
|
|
|
|
2023-10-07 22:34:45 -07:00
|
|
|
|
2021-06-22 16:11:02 -07:00
|
|
|
|
2021-04-16 19:45:58 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-10-06 21:29:08 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-04-16 19:45:58 -07:00
|
|
|
|
2023-10-06 21:29:08 -07:00
|
|
|
|
2021-04-16 19:45:58 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-11-24 19:09:45 -07:00
|
|
|
|
2021-04-16 19:45:58 -07:00
|
|
|
|
2021-06-03 15:39:26 -05:00
|
|
|
|
2021-04-16 19:45:58 -07:00
|
|
|
|
2022-11-27 14:17:55 +00:00
|
|
|
|
2021-04-16 19:45:58 -07:00
|
|
|
|
|
|
|
|
|
2021-10-01 17:45:13 -05:00
|
|
|
|
2021-04-16 19:45:58 -07:00
|
|
|
|
|
|
|
|
|
2023-10-06 21:29:08 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-04-16 19:45:58 -07:00
|
|
|
|
|
|
|
|
|
2022-11-25 17:12:50 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-10-07 22:34:45 -07:00
|
|
|
|
2022-11-25 17:12:50 -07:00
|
|
|
|
2021-04-16 19:45:58 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-11-24 19:09:45 -07:00
|
|
|
|
2021-06-03 15:39:26 -05:00
|
|
|
|
2021-04-16 19:45:58 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-12-30 20:49:02 -07:00
|
|
|
|
2021-04-16 19:45:58 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-10-06 21:29:08 -07:00
|
|
|
|
2022-04-19 21:51:08 -07:00
|
|
|
|
stage2: entry point via std lib and proper updated file detection
Instead of Module setting up the root_scope with the root source file,
instead, Module relies on the package table graph being set up properly,
and inside `update()`, it does the equivalent of `_ = @import("std");`.
This, in term, imports start.zig, which has the logic to call main (or
not). `Module` no longer has `root_scope` - the root source file is no
longer special, it's just in the package table mapped to "root".
I also went ahead and implemented proper detection of updated files.
mtime, inode, size, and source hash are kept in `Scope.File`.
During an update, iterate over `import_table` and stat each file to find
out which ones are updated.
The source hash is redundant with the source hash used by the struct
decl that corresponds to the file, so it should be removed in a future
commit before merging the branch.
* AstGen: add "previously declared here" notes for variables shadowing
decls.
* Parse imports as structs. Module now calls `AstGen.structDeclInner`,
which is called by `AstGen.containerDecl`.
- `importFile` is a bit kludgy with how it handles the top level Decl
that kinda gets merged into the struct decl at the end of the
function. Be on the look out for bugs related to that as well as
possibly cleaner ways to implement this.
* Module: factor out lookupDeclName into lookupIdentifier and lookupNa
* Rename `Scope.Container` to `Scope.Namespace`.
* Delete some dead code.
This branch won't work until `usingnamespace` is implemented because it
relies on `@import("builtin").OutputMode` and `OutputMode` comes from a
`usingnamespace`.
2021-04-09 23:17:50 -07:00
|
|
|
|
2021-04-16 17:28:28 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2022-11-27 14:17:55 +00:00
|
|
|
|
2021-04-16 17:28:28 -07:00
|
|
|
|
stage2: entry point via std lib and proper updated file detection
Instead of Module setting up the root_scope with the root source file,
instead, Module relies on the package table graph being set up properly,
and inside `update()`, it does the equivalent of `_ = @import("std");`.
This, in term, imports start.zig, which has the logic to call main (or
not). `Module` no longer has `root_scope` - the root source file is no
longer special, it's just in the package table mapped to "root".
I also went ahead and implemented proper detection of updated files.
mtime, inode, size, and source hash are kept in `Scope.File`.
During an update, iterate over `import_table` and stat each file to find
out which ones are updated.
The source hash is redundant with the source hash used by the struct
decl that corresponds to the file, so it should be removed in a future
commit before merging the branch.
* AstGen: add "previously declared here" notes for variables shadowing
decls.
* Parse imports as structs. Module now calls `AstGen.structDeclInner`,
which is called by `AstGen.containerDecl`.
- `importFile` is a bit kludgy with how it handles the top level Decl
that kinda gets merged into the struct decl at the end of the
function. Be on the look out for bugs related to that as well as
possibly cleaner ways to implement this.
* Module: factor out lookupDeclName into lookupIdentifier and lookupNa
* Rename `Scope.Container` to `Scope.Namespace`.
* Delete some dead code.
This branch won't work until `usingnamespace` is implemented because it
relies on `@import("builtin").OutputMode` and `OutputMode` comes from a
`usingnamespace`.
2021-04-09 23:17:50 -07:00
|
|
|
|
|
|
|
|
|
2023-02-04 16:14:43 -07:00
|
|
|
|
2021-10-17 18:57:54 -07:00
|
|
|
|
|
|
|
|
|
2023-10-06 21:29:08 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-02-04 16:14:43 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-10-06 21:29:08 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-10-17 18:57:54 -07:00
|
|
|
|
2023-10-06 21:29:08 -07:00
|
|
|
|
2021-10-17 18:57:54 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-11-24 16:58:01 -07:00
|
|
|
|
2021-11-24 19:09:45 -07:00
|
|
|
|
2021-10-17 18:57:54 -07:00
|
|
|
|
2023-10-06 21:29:08 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-10-17 18:57:54 -07:00
|
|
|
|
|
|
|
|
|
2022-11-25 17:12:50 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-10-07 22:34:45 -07:00
|
|
|
|
2022-11-25 17:12:50 -07:00
|
|
|
|
2021-10-17 18:57:54 -07:00
|
|
|
|
|
|
|
|
|
2023-10-06 21:29:08 -07:00
|
|
|
|
2023-02-04 16:14:43 -07:00
|
|
|
|
|
|
|
|
|
2023-10-06 21:29:08 -07:00
|
|
|
|
2023-02-04 16:14:43 -07:00
|
|
|
|
|
|
|
|
|
2023-10-06 21:29:08 -07:00
|
|
|
|
2023-02-04 16:14:43 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-10-06 21:29:08 -07:00
|
|
|
|
2021-10-17 18:57:54 -07:00
|
|
|
|
|
|
|
|
|
2021-12-30 20:49:02 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2022-05-22 19:36:59 +04:30
|
|
|
|
2021-11-21 19:39:32 -07:00
|
|
|
|
2021-11-24 16:58:01 -07:00
|
|
|
|
2021-10-17 18:57:54 -07:00
|
|
|
|
re-apply "Fix C include files not being in `whole` cache (#11365)"
This reverts commit 06310e3d4eb47fed88b175891cb5865bb050f020, reapplying
commit a430630002bf02162ccbf8d3eb10fd73e490cefd.
I deeply apologize to @moosichu and those affected by this bug. The
original fix was actually fine. When I reverted it, I misremembered
how the Cache API works. I thought the fix was going to introduce
nondeterminism into the hash, but I forgot that the order of files in
the manifest doesn't actually matter when checking for a cache hit.
Actually, it does matter a little bit. This fix has a subtle downside
which is that it does introduce the possibility of false negatives when
checking for cache hits of 2+ iterations ago. For example, if the code
goes from "foo", to "bar", and then back to "foo", it may look like a
cache miss when it should have been a hit because 2 iterations ago the
code was the same. However, this is an uncommon use case, and all it
does is cause a bit of wasted time and disk space. That said, my
suggestion from earlier still applies and would be a nice follow-up
enhancement to this fix:
The proper solution to this is to, in whole cache mode, append the hash
inputs to some data structure, and then after the compilation is
complete, do some kind of sorting on the hash inputs so that they will
be the same order every time, then apply them in sequence. No lock on
the Cache object is needed for this scheme.
closes #11063
2022-10-30 16:51:13 -07:00
|
|
|
|
2021-12-30 16:42:32 -07:00
|
|
|
|
|
|
|
|
|
re-apply "Fix C include files not being in `whole` cache (#11365)"
This reverts commit 06310e3d4eb47fed88b175891cb5865bb050f020, reapplying
commit a430630002bf02162ccbf8d3eb10fd73e490cefd.
I deeply apologize to @moosichu and those affected by this bug. The
original fix was actually fine. When I reverted it, I misremembered
how the Cache API works. I thought the fix was going to introduce
nondeterminism into the hash, but I forgot that the order of files in
the manifest doesn't actually matter when checking for a cache hit.
Actually, it does matter a little bit. This fix has a subtle downside
which is that it does introduce the possibility of false negatives when
checking for cache hits of 2+ iterations ago. For example, if the code
goes from "foo", to "bar", and then back to "foo", it may look like a
cache miss when it should have been a hit because 2 iterations ago the
code was the same. However, this is an uncommon use case, and all it
does is cause a bit of wasted time and disk space. That said, my
suggestion from earlier still applies and would be a nice follow-up
enhancement to this fix:
The proper solution to this is to, in whole cache mode, append the hash
inputs to some data structure, and then after the compilation is
complete, do some kind of sorting on the hash inputs so that they will
be the same order every time, then apply them in sequence. No lock on
the Cache object is needed for this scheme.
closes #11063
2022-10-30 16:51:13 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-12-30 16:42:32 -07:00
|
|
|
|
|
|
|
|
|
2023-02-04 16:14:43 -07:00
|
|
|
|
2021-10-17 18:57:54 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-12-30 20:49:02 -07:00
|
|
|
|
2023-10-06 21:29:08 -07:00
|
|
|
|
2021-10-17 18:57:54 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-10-06 21:29:08 -07:00
|
|
|
|
2021-10-17 18:57:54 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-12-30 20:49:02 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-10-17 18:57:54 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2022-05-22 19:36:59 +04:30
|
|
|
|
2021-11-21 19:39:32 -07:00
|
|
|
|
2021-10-17 18:57:54 -07:00
|
|
|
|
|
|
|
|
|
2021-12-30 20:49:02 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-10-17 18:57:54 -07:00
|
|
|
|
2021-11-09 18:27:12 -07:00
|
|
|
|
|
|
|
|
|
2021-10-17 18:57:54 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-04-26 20:41:07 -07:00
|
|
|
|
2021-04-12 16:44:51 -07:00
|
|
|
|
2023-05-08 16:52:59 -07:00
|
|
|
|
2021-04-26 20:41:07 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2022-07-19 01:25:10 +03:00
|
|
|
|
2020-09-13 19:17:58 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-04-26 20:41:07 -07:00
|
|
|
|
2023-05-08 16:52:59 -07:00
|
|
|
|
2021-04-26 20:41:07 -07:00
|
|
|
|
2020-09-13 19:17:58 -07:00
|
|
|
|
2021-04-26 20:41:07 -07:00
|
|
|
|
2021-09-16 18:22:04 -07:00
|
|
|
|
2020-09-13 19:17:58 -07:00
|
|
|
|
2021-04-26 20:41:07 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-04-28 16:55:22 -07:00
|
|
|
|
|
|
|
|
|
2023-05-08 16:52:59 -07:00
|
|
|
|
2021-04-28 16:55:22 -07:00
|
|
|
|
|
|
|
|
|
2021-04-26 20:41:07 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-06-22 18:46:56 +01:00
|
|
|
|
2021-04-28 23:16:13 -07:00
|
|
|
|
|
|
|
|
|
2021-04-28 16:55:22 -07:00
|
|
|
|
2022-01-20 16:08:50 +01:00
|
|
|
|
2023-06-22 18:46:56 +01:00
|
|
|
|
|
|
|
|
|
2021-04-26 20:41:07 -07:00
|
|
|
|
2021-04-28 16:55:22 -07:00
|
|
|
|
2021-04-26 20:41:07 -07:00
|
|
|
|
|
|
|
|
|
2020-09-13 19:17:58 -07:00
|
|
|
|
|
|
|
|
|
2021-04-28 16:55:22 -07:00
|
|
|
|
|
|
|
|
|
2023-05-08 16:52:59 -07:00
|
|
|
|
2021-04-26 20:41:07 -07:00
|
|
|
|
2021-04-28 16:55:22 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2022-07-19 01:25:10 +03:00
|
|
|
|
2021-02-11 23:29:55 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-04-28 16:55:22 -07:00
|
|
|
|
2023-05-08 16:52:59 -07:00
|
|
|
|
|
|
|
|
|
2021-04-26 20:41:07 -07:00
|
|
|
|
|
|
|
|
|
compiler: eliminate Decl.value_arena and Sema.perm_arena
The main motivation for this commit is eliminating Decl.value_arena.
Everything else is dominoes.
Decl.name used to be stored in the GPA, now it is stored in InternPool.
It ended up being simpler to migrate other strings to be interned as
well, such as struct field names, union field names, and a few others.
This ended up requiring a big diff, sorry about that. But the changes
are pretty nice, we finally start to take advantage of InternPool's
existence.
global_error_set and error_name_list are simplified. Now it is a single
ArrayHashMap(NullTerminatedString, void) and the index is the error tag
value.
Module.tmp_hack_arena is re-introduced (it was removed in
eeff407941560ce8eb5b737b2436dfa93cfd3a0c) in order to deal with
comptime_args, optimized_order, and struct and union fields. After
structs and unions get moved into InternPool properly, tmp_hack_arena
can be deleted again.
2023-06-01 23:46:04 -07:00
|
|
|
|
2021-04-27 18:36:12 -07:00
|
|
|
|
2021-04-28 16:55:22 -07:00
|
|
|
|
2021-09-02 14:46:31 +02:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-04-28 16:55:22 -07:00
|
|
|
|
2021-04-26 20:41:07 -07:00
|
|
|
|
2022-01-31 17:07:45 +01:00
|
|
|
|
|
|
|
|
|
2021-05-01 21:57:52 -07:00
|
|
|
|
2022-01-27 15:23:28 -05:00
|
|
|
|
2022-04-19 21:51:08 -07:00
|
|
|
|
|
|
|
|
|
2021-04-28 16:55:22 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-05-07 18:52:11 -07:00
|
|
|
|
2023-01-10 21:54:58 +02:00
|
|
|
|
compiler: eliminate Decl.value_arena and Sema.perm_arena
The main motivation for this commit is eliminating Decl.value_arena.
Everything else is dominoes.
Decl.name used to be stored in the GPA, now it is stored in InternPool.
It ended up being simpler to migrate other strings to be interned as
well, such as struct field names, union field names, and a few others.
This ended up requiring a big diff, sorry about that. But the changes
are pretty nice, we finally start to take advantage of InternPool's
existence.
global_error_set and error_name_list are simplified. Now it is a single
ArrayHashMap(NullTerminatedString, void) and the index is the error tag
value.
Module.tmp_hack_arena is re-introduced (it was removed in
eeff407941560ce8eb5b737b2436dfa93cfd3a0c) in order to deal with
comptime_args, optimized_order, and struct and union fields. After
structs and unions get moved into InternPool properly, tmp_hack_arena
can be deleted again.
2023-06-01 23:46:04 -07:00
|
|
|
|
2021-04-28 16:55:22 -07:00
|
|
|
|
2021-08-28 15:35:59 -07:00
|
|
|
|
2021-04-28 16:55:22 -07:00
|
|
|
|
|
|
|
|
|
2023-01-10 21:54:58 +02:00
|
|
|
|
compiler: eliminate Decl.value_arena and Sema.perm_arena
The main motivation for this commit is eliminating Decl.value_arena.
Everything else is dominoes.
Decl.name used to be stored in the GPA, now it is stored in InternPool.
It ended up being simpler to migrate other strings to be interned as
well, such as struct field names, union field names, and a few others.
This ended up requiring a big diff, sorry about that. But the changes
are pretty nice, we finally start to take advantage of InternPool's
existence.
global_error_set and error_name_list are simplified. Now it is a single
ArrayHashMap(NullTerminatedString, void) and the index is the error tag
value.
Module.tmp_hack_arena is re-introduced (it was removed in
eeff407941560ce8eb5b737b2436dfa93cfd3a0c) in order to deal with
comptime_args, optimized_order, and struct and union fields. After
structs and unions get moved into InternPool properly, tmp_hack_arena
can be deleted again.
2023-06-01 23:46:04 -07:00
|
|
|
|
2021-04-28 16:55:22 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-01-10 21:54:58 +02:00
|
|
|
|
compiler: eliminate Decl.value_arena and Sema.perm_arena
The main motivation for this commit is eliminating Decl.value_arena.
Everything else is dominoes.
Decl.name used to be stored in the GPA, now it is stored in InternPool.
It ended up being simpler to migrate other strings to be interned as
well, such as struct field names, union field names, and a few others.
This ended up requiring a big diff, sorry about that. But the changes
are pretty nice, we finally start to take advantage of InternPool's
existence.
global_error_set and error_name_list are simplified. Now it is a single
ArrayHashMap(NullTerminatedString, void) and the index is the error tag
value.
Module.tmp_hack_arena is re-introduced (it was removed in
eeff407941560ce8eb5b737b2436dfa93cfd3a0c) in order to deal with
comptime_args, optimized_order, and struct and union fields. After
structs and unions get moved into InternPool properly, tmp_hack_arena
can be deleted again.
2023-06-01 23:46:04 -07:00
|
|
|
|
2021-04-28 16:55:22 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-01-10 21:54:58 +02:00
|
|
|
|
compiler: eliminate Decl.value_arena and Sema.perm_arena
The main motivation for this commit is eliminating Decl.value_arena.
Everything else is dominoes.
Decl.name used to be stored in the GPA, now it is stored in InternPool.
It ended up being simpler to migrate other strings to be interned as
well, such as struct field names, union field names, and a few others.
This ended up requiring a big diff, sorry about that. But the changes
are pretty nice, we finally start to take advantage of InternPool's
existence.
global_error_set and error_name_list are simplified. Now it is a single
ArrayHashMap(NullTerminatedString, void) and the index is the error tag
value.
Module.tmp_hack_arena is re-introduced (it was removed in
eeff407941560ce8eb5b737b2436dfa93cfd3a0c) in order to deal with
comptime_args, optimized_order, and struct and union fields. After
structs and unions get moved into InternPool properly, tmp_hack_arena
can be deleted again.
2023-06-01 23:46:04 -07:00
|
|
|
|
2021-04-28 16:55:22 -07:00
|
|
|
|
2022-01-27 15:23:28 -05:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-01-10 21:54:58 +02:00
|
|
|
|
compiler: eliminate Decl.value_arena and Sema.perm_arena
The main motivation for this commit is eliminating Decl.value_arena.
Everything else is dominoes.
Decl.name used to be stored in the GPA, now it is stored in InternPool.
It ended up being simpler to migrate other strings to be interned as
well, such as struct field names, union field names, and a few others.
This ended up requiring a big diff, sorry about that. But the changes
are pretty nice, we finally start to take advantage of InternPool's
existence.
global_error_set and error_name_list are simplified. Now it is a single
ArrayHashMap(NullTerminatedString, void) and the index is the error tag
value.
Module.tmp_hack_arena is re-introduced (it was removed in
eeff407941560ce8eb5b737b2436dfa93cfd3a0c) in order to deal with
comptime_args, optimized_order, and struct and union fields. After
structs and unions get moved into InternPool properly, tmp_hack_arena
can be deleted again.
2023-06-01 23:46:04 -07:00
|
|
|
|
2022-01-27 15:23:28 -05:00
|
|
|
|
2021-05-07 18:52:11 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-01-10 21:54:58 +02:00
|
|
|
|
compiler: eliminate Decl.value_arena and Sema.perm_arena
The main motivation for this commit is eliminating Decl.value_arena.
Everything else is dominoes.
Decl.name used to be stored in the GPA, now it is stored in InternPool.
It ended up being simpler to migrate other strings to be interned as
well, such as struct field names, union field names, and a few others.
This ended up requiring a big diff, sorry about that. But the changes
are pretty nice, we finally start to take advantage of InternPool's
existence.
global_error_set and error_name_list are simplified. Now it is a single
ArrayHashMap(NullTerminatedString, void) and the index is the error tag
value.
Module.tmp_hack_arena is re-introduced (it was removed in
eeff407941560ce8eb5b737b2436dfa93cfd3a0c) in order to deal with
comptime_args, optimized_order, and struct and union fields. After
structs and unions get moved into InternPool properly, tmp_hack_arena
can be deleted again.
2023-06-01 23:46:04 -07:00
|
|
|
|
2021-05-07 18:52:11 -07:00
|
|
|
|
compiler: eliminate Decl.value_arena and Sema.perm_arena
The main motivation for this commit is eliminating Decl.value_arena.
Everything else is dominoes.
Decl.name used to be stored in the GPA, now it is stored in InternPool.
It ended up being simpler to migrate other strings to be interned as
well, such as struct field names, union field names, and a few others.
This ended up requiring a big diff, sorry about that. But the changes
are pretty nice, we finally start to take advantage of InternPool's
existence.
global_error_set and error_name_list are simplified. Now it is a single
ArrayHashMap(NullTerminatedString, void) and the index is the error tag
value.
Module.tmp_hack_arena is re-introduced (it was removed in
eeff407941560ce8eb5b737b2436dfa93cfd3a0c) in order to deal with
comptime_args, optimized_order, and struct and union fields. After
structs and unions get moved into InternPool properly, tmp_hack_arena
can be deleted again.
2023-06-01 23:46:04 -07:00
|
|
|
|
2021-05-07 18:52:11 -07:00
|
|
|
|
|
|
|
|
|
2021-05-02 14:58:27 -07:00
|
|
|
|
2023-05-08 17:59:06 +10:00
|
|
|
|
2021-08-28 15:35:59 -07:00
|
|
|
|
2023-01-10 21:54:58 +02:00
|
|
|
|
2021-04-26 21:34:40 -07:00
|
|
|
|
2021-04-26 20:41:07 -07:00
|
|
|
|
2022-04-19 21:51:08 -07:00
|
|
|
|
|
|
|
|
|
compiler: eliminate Decl.value_arena and Sema.perm_arena
The main motivation for this commit is eliminating Decl.value_arena.
Everything else is dominoes.
Decl.name used to be stored in the GPA, now it is stored in InternPool.
It ended up being simpler to migrate other strings to be interned as
well, such as struct field names, union field names, and a few others.
This ended up requiring a big diff, sorry about that. But the changes
are pretty nice, we finally start to take advantage of InternPool's
existence.
global_error_set and error_name_list are simplified. Now it is a single
ArrayHashMap(NullTerminatedString, void) and the index is the error tag
value.
Module.tmp_hack_arena is re-introduced (it was removed in
eeff407941560ce8eb5b737b2436dfa93cfd3a0c) in order to deal with
comptime_args, optimized_order, and struct and union fields. After
structs and unions get moved into InternPool properly, tmp_hack_arena
can be deleted again.
2023-06-01 23:46:04 -07:00
|
|
|
|
2022-04-19 21:51:08 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2022-06-27 14:31:54 -07:00
|
|
|
|
2021-04-26 20:41:07 -07:00
|
|
|
|
2023-05-08 16:52:59 -07:00
|
|
|
|
2022-04-19 21:51:08 -07:00
|
|
|
|
2023-01-10 21:54:58 +02:00
|
|
|
|
2022-04-19 21:51:08 -07:00
|
|
|
|
2023-01-10 21:54:58 +02:00
|
|
|
|
2022-04-19 21:51:08 -07:00
|
|
|
|
2021-08-28 15:35:59 -07:00
|
|
|
|
2021-05-01 21:57:52 -07:00
|
|
|
|
2022-04-19 21:51:08 -07:00
|
|
|
|
2021-04-26 21:34:40 -07:00
|
|
|
|
|
|
|
|
|
2023-10-06 21:29:08 -07:00
|
|
|
|
2021-04-26 21:34:40 -07:00
|
|
|
|
2021-08-28 15:35:59 -07:00
|
|
|
|
2021-07-23 22:23:03 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2022-06-27 14:31:54 -07:00
|
|
|
|
2023-10-06 21:29:08 -07:00
|
|
|
|
2022-04-19 21:51:08 -07:00
|
|
|
|
2021-07-23 22:23:03 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2022-06-27 14:31:54 -07:00
|
|
|
|
2023-10-06 21:29:08 -07:00
|
|
|
|
2022-06-27 14:31:54 -07:00
|
|
|
|
compiler: eliminate Decl.value_arena and Sema.perm_arena
The main motivation for this commit is eliminating Decl.value_arena.
Everything else is dominoes.
Decl.name used to be stored in the GPA, now it is stored in InternPool.
It ended up being simpler to migrate other strings to be interned as
well, such as struct field names, union field names, and a few others.
This ended up requiring a big diff, sorry about that. But the changes
are pretty nice, we finally start to take advantage of InternPool's
existence.
global_error_set and error_name_list are simplified. Now it is a single
ArrayHashMap(NullTerminatedString, void) and the index is the error tag
value.
Module.tmp_hack_arena is re-introduced (it was removed in
eeff407941560ce8eb5b737b2436dfa93cfd3a0c) in order to deal with
comptime_args, optimized_order, and struct and union fields. After
structs and unions get moved into InternPool properly, tmp_hack_arena
can be deleted again.
2023-06-01 23:46:04 -07:00
|
|
|
|
2022-06-27 14:31:54 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2022-04-19 21:51:08 -07:00
|
|
|
|
2021-07-23 22:23:03 -07:00
|
|
|
|
|
|
|
|
|
2021-04-26 21:34:40 -07:00
|
|
|
|
|
|
|
|
|
2022-06-27 14:31:54 -07:00
|
|
|
|
2021-04-26 20:41:07 -07:00
|
|
|
|
|
|
|
|
|
2021-04-28 16:55:22 -07:00
|
|
|
|
|
|
|
|
|
2021-09-02 14:46:31 +02:00
|
|
|
|
2023-06-22 18:46:56 +01:00
|
|
|
|
stage2: garbage collect unused anon decls
After this change, the frontend and backend cooperate to keep track of
which Decls are actually emitted into the machine code. When any backend
sees a `decl_ref` Value, it must mark the corresponding Decl `alive`
field to true.
This prevents unused comptime data from spilling into the output object
files. For example, if you do an `inline for` loop, previously, any
intermediate value calculations would have gone into the object file.
Now they are garbage collected immediately after the owner Decl has its
machine code generated.
In the frontend, when it is time to send a Decl to the linker, if it has
not been marked "alive" then it is deleted instead.
Additional improvements:
* Resolve type ABI layouts after successful semantic analysis of a
Decl. This is needed so that the backend has access to struct fields.
* Sema: fix incorrect logic in resolveMaybeUndefVal. It should return
"not comptime known" instead of a compile error for global variables.
* `Value.pointerDeref` now returns `null` in the case that the pointer
deref cannot happen at compile-time. This is true for global
variables, for example. Another example is if a comptime known
pointer has a hard coded address value.
* Binary arithmetic sets the requireRuntimeBlock source location to the
lhs_src or rhs_src as appropriate instead of on the operator node.
* Fix LLVM codegen for slice_elem_val which had the wrong logic for
when the operand was not a pointer.
As noted in the comment in the implementation of deleteUnusedDecl, a
future improvement will be to rework the frontend/linker interface to
remove the frontend's responsibility of calling allocateDeclIndexes.
I discovered some issues with the plan9 linker backend that are related
to this, and worked around them for now.
2021-07-29 19:30:37 -07:00
|
|
|
|
2021-04-26 20:41:07 -07:00
|
|
|
|
2021-02-11 23:29:55 -07:00
|
|
|
|
2022-04-19 21:51:08 -07:00
|
|
|
|
|
|
|
|
|
2023-05-08 17:59:06 +10:00
|
|
|
|
|
|
|
|
|
2023-05-08 16:52:59 -07:00
|
|
|
|
2023-05-08 17:59:06 +10:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-06-12 01:44:12 -04:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-05-08 17:59:06 +10:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-04-26 20:41:07 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-05-01 21:57:52 -07:00
|
|
|
|
2021-04-28 16:55:22 -07:00
|
|
|
|
2021-04-27 18:36:12 -07:00
|
|
|
|
|
|
|
|
|
2023-01-10 21:54:58 +02:00
|
|
|
|
2021-04-28 16:55:22 -07:00
|
|
|
|
2021-09-02 14:46:31 +02:00
|
|
|
|
2023-06-22 18:46:56 +01:00
|
|
|
|
2023-07-17 17:47:59 -07:00
|
|
|
|
2022-06-27 14:31:54 -07:00
|
|
|
|
2023-02-01 16:01:43 +01:00
|
|
|
|
2021-08-29 15:11:01 -04:00
|
|
|
|
|
|
|
|
|
2022-06-27 14:31:54 -07:00
|
|
|
|
2021-06-01 16:07:08 -04:00
|
|
|
|
2022-02-05 15:33:00 +01:00
|
|
|
|
2021-05-06 17:20:45 -07:00
|
|
|
|
2021-02-11 23:29:55 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
stage2: fix deletion of Decls that get re-referenced
When scanDecls happens, we create stub Decl objects that
have not been semantically analyzed. When they get referenced,
they get semantically analyzed.
Before this commit, when they got unreferenced, they were completely
deleted, including deleted from the containing Namespace.
However, if the update did not cause the containing Namespace to get
deleted, for example, if `std.builtin.ExportOptions` is no longer
referenced, but `std.builtin` is still referenced, and then `ExportOptions`
gets referenced again, the Namespace would be incorrectly missing the
Decl, so we get an incorrect "no such member" error.
The solution is to, when dealing with a no longer referenced Decl
objects during an update, clear them to the state they would be in
on a fresh scanDecl, rather than completely deleting them.
2021-05-18 12:35:36 -07:00
|
|
|
|
|
|
|
|
|
2021-04-07 19:38:00 -07:00
|
|
|
|
2022-04-19 21:51:08 -07:00
|
|
|
|
|
|
|
|
|
2021-05-11 22:12:36 -07:00
|
|
|
|
2021-02-11 23:29:55 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2022-04-19 21:51:08 -07:00
|
|
|
|
2021-04-07 19:38:00 -07:00
|
|
|
|
2021-05-11 22:12:36 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-04-07 19:38:00 -07:00
|
|
|
|
2022-04-19 21:51:08 -07:00
|
|
|
|
2021-05-07 18:52:11 -07:00
|
|
|
|
2021-04-07 19:38:00 -07:00
|
|
|
|
2020-09-13 19:17:58 -07:00
|
|
|
|
stage2: fix deletion of Decls that get re-referenced
When scanDecls happens, we create stub Decl objects that
have not been semantically analyzed. When they get referenced,
they get semantically analyzed.
Before this commit, when they got unreferenced, they were completely
deleted, including deleted from the containing Namespace.
However, if the update did not cause the containing Namespace to get
deleted, for example, if `std.builtin.ExportOptions` is no longer
referenced, but `std.builtin` is still referenced, and then `ExportOptions`
gets referenced again, the Namespace would be incorrectly missing the
Decl, so we get an incorrect "no such member" error.
The solution is to, when dealing with a no longer referenced Decl
objects during an update, clear them to the state they would be in
on a fresh scanDecl, rather than completely deleting them.
2021-05-18 12:35:36 -07:00
|
|
|
|
2022-04-19 21:51:08 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-06-03 15:39:26 -05:00
|
|
|
|
2020-09-13 19:17:58 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2022-04-19 21:51:08 -07:00
|
|
|
|
2020-09-13 19:17:58 -07:00
|
|
|
|
|
|
|
|
|
stage2: fix deletion of Decls that get re-referenced
When scanDecls happens, we create stub Decl objects that
have not been semantically analyzed. When they get referenced,
they get semantically analyzed.
Before this commit, when they got unreferenced, they were completely
deleted, including deleted from the containing Namespace.
However, if the update did not cause the containing Namespace to get
deleted, for example, if `std.builtin.ExportOptions` is no longer
referenced, but `std.builtin` is still referenced, and then `ExportOptions`
gets referenced again, the Namespace would be incorrectly missing the
Decl, so we get an incorrect "no such member" error.
The solution is to, when dealing with a no longer referenced Decl
objects during an update, clear them to the state they would be in
on a fresh scanDecl, rather than completely deleting them.
2021-05-18 12:35:36 -07:00
|
|
|
|
|
|
|
|
|
2021-04-07 19:38:00 -07:00
|
|
|
|
2022-04-19 21:51:08 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-04-07 19:38:00 -07:00
|
|
|
|
2022-04-19 21:51:08 -07:00
|
|
|
|
2020-09-13 19:17:58 -07:00
|
|
|
|
|
|
|
|
|
stage2: fix deletion of Decls that get re-referenced
When scanDecls happens, we create stub Decl objects that
have not been semantically analyzed. When they get referenced,
they get semantically analyzed.
Before this commit, when they got unreferenced, they were completely
deleted, including deleted from the containing Namespace.
However, if the update did not cause the containing Namespace to get
deleted, for example, if `std.builtin.ExportOptions` is no longer
referenced, but `std.builtin` is still referenced, and then `ExportOptions`
gets referenced again, the Namespace would be incorrectly missing the
Decl, so we get an incorrect "no such member" error.
The solution is to, when dealing with a no longer referenced Decl
objects during an update, clear them to the state they would be in
on a fresh scanDecl, rather than completely deleting them.
2021-05-18 12:35:36 -07:00
|
|
|
|
|
|
|
|
|
2022-04-19 21:51:08 -07:00
|
|
|
|
2021-06-03 15:39:26 -05:00
|
|
|
|
2020-09-13 19:17:58 -07:00
|
|
|
|
2022-12-12 13:16:09 +02:00
|
|
|
|
2023-09-25 04:54:07 +02:00
|
|
|
|
|
|
|
|
|
2022-12-12 13:16:09 +02:00
|
|
|
|
2021-04-26 20:41:07 -07:00
|
|
|
|
2022-04-19 21:51:08 -07:00
|
|
|
|
2021-06-03 15:39:26 -05:00
|
|
|
|
2021-04-26 20:41:07 -07:00
|
|
|
|
2022-04-19 21:51:08 -07:00
|
|
|
|
2021-01-05 17:33:31 -07:00
|
|
|
|
2022-04-19 21:51:08 -07:00
|
|
|
|
2023-01-30 18:22:50 +01:00
|
|
|
|
2021-05-11 22:12:36 -07:00
|
|
|
|
|
|
|
|
|
2023-05-02 15:01:45 -07:00
|
|
|
|
2022-04-19 21:51:08 -07:00
|
|
|
|
2021-05-17 13:44:20 -07:00
|
|
|
|
2023-05-26 16:04:53 -04:00
|
|
|
|
2021-05-11 22:12:36 -07:00
|
|
|
|
|
|
|
|
|
2021-05-07 18:52:11 -07:00
|
|
|
|
2021-01-05 17:33:31 -07:00
|
|
|
|
stage2: fix deletion of Decls that get re-referenced
When scanDecls happens, we create stub Decl objects that
have not been semantically analyzed. When they get referenced,
they get semantically analyzed.
Before this commit, when they got unreferenced, they were completely
deleted, including deleted from the containing Namespace.
However, if the update did not cause the containing Namespace to get
deleted, for example, if `std.builtin.ExportOptions` is no longer
referenced, but `std.builtin` is still referenced, and then `ExportOptions`
gets referenced again, the Namespace would be incorrectly missing the
Decl, so we get an incorrect "no such member" error.
The solution is to, when dealing with a no longer referenced Decl
objects during an update, clear them to the state they would be in
on a fresh scanDecl, rather than completely deleting them.
2021-05-18 12:35:36 -07:00
|
|
|
|
|
|
|
|
|
2022-04-19 21:51:08 -07:00
|
|
|
|
stage2: fix deletion of Decls that get re-referenced
When scanDecls happens, we create stub Decl objects that
have not been semantically analyzed. When they get referenced,
they get semantically analyzed.
Before this commit, when they got unreferenced, they were completely
deleted, including deleted from the containing Namespace.
However, if the update did not cause the containing Namespace to get
deleted, for example, if `std.builtin.ExportOptions` is no longer
referenced, but `std.builtin` is still referenced, and then `ExportOptions`
gets referenced again, the Namespace would be incorrectly missing the
Decl, so we get an incorrect "no such member" error.
The solution is to, when dealing with a no longer referenced Decl
objects during an update, clear them to the state they would be in
on a fresh scanDecl, rather than completely deleting them.
2021-05-18 12:35:36 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2020-09-13 19:17:58 -07:00
|
|
|
|
|
|
|
|
|
stage2: implement runtime pointer access to global constants
The main problem that motivated these changes is that global constants
which are referenced by pointer would not be emitted into the binary.
This happened because `semaDecl` did not add `codegen_decl` tasks for
global constants, instead relying on the constant values being copied as
necessary. However when the global constants are referenced by pointer,
they need to be sent to the linker to be emitted.
After making global const arrays, structs, and unions get emitted, this
uncovered a latent issue: the anonymous decls that they referenced would
get garbage collected (via `deleteUnusedDecl`) even though they would
later be referenced by the global const.
In order to solve this problem, I introduced `anon_work_queue` which is
the same as `work_queue` except a lower priority. The `codegen_decl`
task for anon decls goes into the `anon_work_queue` ensuring that the
owner decl gets a chance to mark its anon decls as alive before they are
possibly deleted.
This caused a few regressions, which I made the judgement call to add
workarounds for. Two steps forward, one step back, is still progress.
The regressions were:
* Two behavior tests having to do with unions. These tests were
intentionally exercising the LLVM constant value lowering, however,
due to the bug with garbage collection that was fixed in this commit,
the LLVM code was not getting exercised, and union types/values were
not implemented correctly, due to me forgetting that LLVM does not
allow bitcasting aggregate values.
- This is worked around by allowing those 2 test cases to regress,
moving them to the "passing for stage1 only" section.
* The test-stage2 test cases (in test/cases/*) for non-LLVM backends
previously did not have any calls to lower struct values, but now
they do. The code that was there was just `@panic("TODO")`. I
replaced that code with a stub that generates the wrong value. This
is an intentional miscompilation that will obviously need to get
fixed before any struct behavior tests pass. None of the current
tests we have exercise loading any values from these global const
structs, so there is not a problem until we try to improve these
backends.
2021-10-26 22:29:43 -07:00
|
|
|
|
2023-07-06 15:27:47 -07:00
|
|
|
|
|
|
|
|
|
2022-04-19 21:51:08 -07:00
|
|
|
|
2023-07-06 15:27:47 -07:00
|
|
|
|
|
|
|
|
|
stage2: garbage collect unused anon decls
After this change, the frontend and backend cooperate to keep track of
which Decls are actually emitted into the machine code. When any backend
sees a `decl_ref` Value, it must mark the corresponding Decl `alive`
field to true.
This prevents unused comptime data from spilling into the output object
files. For example, if you do an `inline for` loop, previously, any
intermediate value calculations would have gone into the object file.
Now they are garbage collected immediately after the owner Decl has its
machine code generated.
In the frontend, when it is time to send a Decl to the linker, if it has
not been marked "alive" then it is deleted instead.
Additional improvements:
* Resolve type ABI layouts after successful semantic analysis of a
Decl. This is needed so that the backend has access to struct fields.
* Sema: fix incorrect logic in resolveMaybeUndefVal. It should return
"not comptime known" instead of a compile error for global variables.
* `Value.pointerDeref` now returns `null` in the case that the pointer
deref cannot happen at compile-time. This is true for global
variables, for example. Another example is if a comptime known
pointer has a hard coded address value.
* Binary arithmetic sets the requireRuntimeBlock source location to the
lhs_src or rhs_src as appropriate instead of on the operator node.
* Fix LLVM codegen for slice_elem_val which had the wrong logic for
when the operand was not a pointer.
As noted in the comment in the implementation of deleteUnusedDecl, a
future improvement will be to rework the frontend/linker interface to
remove the frontend's responsibility of calling allocateDeclIndexes.
I discovered some issues with the plan9 linker backend that are related
to this, and worked around them for now.
2021-07-29 19:30:37 -07:00
|
|
|
|
2023-07-06 15:27:47 -07:00
|
|
|
|
stage2: garbage collect unused anon decls
After this change, the frontend and backend cooperate to keep track of
which Decls are actually emitted into the machine code. When any backend
sees a `decl_ref` Value, it must mark the corresponding Decl `alive`
field to true.
This prevents unused comptime data from spilling into the output object
files. For example, if you do an `inline for` loop, previously, any
intermediate value calculations would have gone into the object file.
Now they are garbage collected immediately after the owner Decl has its
machine code generated.
In the frontend, when it is time to send a Decl to the linker, if it has
not been marked "alive" then it is deleted instead.
Additional improvements:
* Resolve type ABI layouts after successful semantic analysis of a
Decl. This is needed so that the backend has access to struct fields.
* Sema: fix incorrect logic in resolveMaybeUndefVal. It should return
"not comptime known" instead of a compile error for global variables.
* `Value.pointerDeref` now returns `null` in the case that the pointer
deref cannot happen at compile-time. This is true for global
variables, for example. Another example is if a comptime known
pointer has a hard coded address value.
* Binary arithmetic sets the requireRuntimeBlock source location to the
lhs_src or rhs_src as appropriate instead of on the operator node.
* Fix LLVM codegen for slice_elem_val which had the wrong logic for
when the operand was not a pointer.
As noted in the comment in the implementation of deleteUnusedDecl, a
future improvement will be to rework the frontend/linker interface to
remove the frontend's responsibility of calling allocateDeclIndexes.
I discovered some issues with the plan9 linker backend that are related
to this, and worked around them for now.
2021-07-29 19:30:37 -07:00
|
|
|
|
2023-07-06 15:27:47 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
stage2: garbage collect unused anon decls
After this change, the frontend and backend cooperate to keep track of
which Decls are actually emitted into the machine code. When any backend
sees a `decl_ref` Value, it must mark the corresponding Decl `alive`
field to true.
This prevents unused comptime data from spilling into the output object
files. For example, if you do an `inline for` loop, previously, any
intermediate value calculations would have gone into the object file.
Now they are garbage collected immediately after the owner Decl has its
machine code generated.
In the frontend, when it is time to send a Decl to the linker, if it has
not been marked "alive" then it is deleted instead.
Additional improvements:
* Resolve type ABI layouts after successful semantic analysis of a
Decl. This is needed so that the backend has access to struct fields.
* Sema: fix incorrect logic in resolveMaybeUndefVal. It should return
"not comptime known" instead of a compile error for global variables.
* `Value.pointerDeref` now returns `null` in the case that the pointer
deref cannot happen at compile-time. This is true for global
variables, for example. Another example is if a comptime known
pointer has a hard coded address value.
* Binary arithmetic sets the requireRuntimeBlock source location to the
lhs_src or rhs_src as appropriate instead of on the operator node.
* Fix LLVM codegen for slice_elem_val which had the wrong logic for
when the operand was not a pointer.
As noted in the comment in the implementation of deleteUnusedDecl, a
future improvement will be to rework the frontend/linker interface to
remove the frontend's responsibility of calling allocateDeclIndexes.
I discovered some issues with the plan9 linker backend that are related
to this, and worked around them for now.
2021-07-29 19:30:37 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
stage2: implement runtime pointer access to global constants
The main problem that motivated these changes is that global constants
which are referenced by pointer would not be emitted into the binary.
This happened because `semaDecl` did not add `codegen_decl` tasks for
global constants, instead relying on the constant values being copied as
necessary. However when the global constants are referenced by pointer,
they need to be sent to the linker to be emitted.
After making global const arrays, structs, and unions get emitted, this
uncovered a latent issue: the anonymous decls that they referenced would
get garbage collected (via `deleteUnusedDecl`) even though they would
later be referenced by the global const.
In order to solve this problem, I introduced `anon_work_queue` which is
the same as `work_queue` except a lower priority. The `codegen_decl`
task for anon decls goes into the `anon_work_queue` ensuring that the
owner decl gets a chance to mark its anon decls as alive before they are
possibly deleted.
This caused a few regressions, which I made the judgement call to add
workarounds for. Two steps forward, one step back, is still progress.
The regressions were:
* Two behavior tests having to do with unions. These tests were
intentionally exercising the LLVM constant value lowering, however,
due to the bug with garbage collection that was fixed in this commit,
the LLVM code was not getting exercised, and union types/values were
not implemented correctly, due to me forgetting that LLVM does not
allow bitcasting aggregate values.
- This is worked around by allowing those 2 test cases to regress,
moving them to the "passing for stage1 only" section.
* The test-stage2 test cases (in test/cases/*) for non-LLVM backends
previously did not have any calls to lower struct values, but now
they do. The code that was there was just `@panic("TODO")`. I
replaced that code with a stub that generates the wrong value. This
is an intentional miscompilation that will obviously need to get
fixed before any struct behavior tests pass. None of the current
tests we have exercise loading any values from these global const
structs, so there is not a problem until we try to improve these
backends.
2021-10-26 22:29:43 -07:00
|
|
|
|
|
|
|
|
|
2022-04-19 21:51:08 -07:00
|
|
|
|
|
|
|
|
|
stage2: implement runtime pointer access to global constants
The main problem that motivated these changes is that global constants
which are referenced by pointer would not be emitted into the binary.
This happened because `semaDecl` did not add `codegen_decl` tasks for
global constants, instead relying on the constant values being copied as
necessary. However when the global constants are referenced by pointer,
they need to be sent to the linker to be emitted.
After making global const arrays, structs, and unions get emitted, this
uncovered a latent issue: the anonymous decls that they referenced would
get garbage collected (via `deleteUnusedDecl`) even though they would
later be referenced by the global const.
In order to solve this problem, I introduced `anon_work_queue` which is
the same as `work_queue` except a lower priority. The `codegen_decl`
task for anon decls goes into the `anon_work_queue` ensuring that the
owner decl gets a chance to mark its anon decls as alive before they are
possibly deleted.
This caused a few regressions, which I made the judgement call to add
workarounds for. Two steps forward, one step back, is still progress.
The regressions were:
* Two behavior tests having to do with unions. These tests were
intentionally exercising the LLVM constant value lowering, however,
due to the bug with garbage collection that was fixed in this commit,
the LLVM code was not getting exercised, and union types/values were
not implemented correctly, due to me forgetting that LLVM does not
allow bitcasting aggregate values.
- This is worked around by allowing those 2 test cases to regress,
moving them to the "passing for stage1 only" section.
* The test-stage2 test cases (in test/cases/*) for non-LLVM backends
previously did not have any calls to lower struct values, but now
they do. The code that was there was just `@panic("TODO")`. I
replaced that code with a stub that generates the wrong value. This
is an intentional miscompilation that will obviously need to get
fixed before any struct behavior tests pass. None of the current
tests we have exercise loading any values from these global const
structs, so there is not a problem until we try to improve these
backends.
2021-10-26 22:29:43 -07:00
|
|
|
|
2022-04-19 21:51:08 -07:00
|
|
|
|
stage2: implement runtime pointer access to global constants
The main problem that motivated these changes is that global constants
which are referenced by pointer would not be emitted into the binary.
This happened because `semaDecl` did not add `codegen_decl` tasks for
global constants, instead relying on the constant values being copied as
necessary. However when the global constants are referenced by pointer,
they need to be sent to the linker to be emitted.
After making global const arrays, structs, and unions get emitted, this
uncovered a latent issue: the anonymous decls that they referenced would
get garbage collected (via `deleteUnusedDecl`) even though they would
later be referenced by the global const.
In order to solve this problem, I introduced `anon_work_queue` which is
the same as `work_queue` except a lower priority. The `codegen_decl`
task for anon decls goes into the `anon_work_queue` ensuring that the
owner decl gets a chance to mark its anon decls as alive before they are
possibly deleted.
This caused a few regressions, which I made the judgement call to add
workarounds for. Two steps forward, one step back, is still progress.
The regressions were:
* Two behavior tests having to do with unions. These tests were
intentionally exercising the LLVM constant value lowering, however,
due to the bug with garbage collection that was fixed in this commit,
the LLVM code was not getting exercised, and union types/values were
not implemented correctly, due to me forgetting that LLVM does not
allow bitcasting aggregate values.
- This is worked around by allowing those 2 test cases to regress,
moving them to the "passing for stage1 only" section.
* The test-stage2 test cases (in test/cases/*) for non-LLVM backends
previously did not have any calls to lower struct values, but now
they do. The code that was there was just `@panic("TODO")`. I
replaced that code with a stub that generates the wrong value. This
is an intentional miscompilation that will obviously need to get
fixed before any struct behavior tests pass. None of the current
tests we have exercise loading any values from these global const
structs, so there is not a problem until we try to improve these
backends.
2021-10-26 22:29:43 -07:00
|
|
|
|
|
|
|
|
|
2021-10-01 12:45:46 -05:00
|
|
|
|
|
|
|
|
|
2022-04-19 21:51:08 -07:00
|
|
|
|
|
|
|
|
|
2021-09-30 23:38:41 -05:00
|
|
|
|
2022-04-19 21:51:08 -07:00
|
|
|
|
2021-09-30 23:38:41 -05:00
|
|
|
|
2021-10-01 12:45:46 -05:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-09-30 23:38:41 -05:00
|
|
|
|
2022-04-19 21:51:08 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-09-30 23:38:41 -05:00
|
|
|
|
|
|
|
|
|
2022-04-19 21:51:08 -07:00
|
|
|
|
stage2: garbage collect unused anon decls
After this change, the frontend and backend cooperate to keep track of
which Decls are actually emitted into the machine code. When any backend
sees a `decl_ref` Value, it must mark the corresponding Decl `alive`
field to true.
This prevents unused comptime data from spilling into the output object
files. For example, if you do an `inline for` loop, previously, any
intermediate value calculations would have gone into the object file.
Now they are garbage collected immediately after the owner Decl has its
machine code generated.
In the frontend, when it is time to send a Decl to the linker, if it has
not been marked "alive" then it is deleted instead.
Additional improvements:
* Resolve type ABI layouts after successful semantic analysis of a
Decl. This is needed so that the backend has access to struct fields.
* Sema: fix incorrect logic in resolveMaybeUndefVal. It should return
"not comptime known" instead of a compile error for global variables.
* `Value.pointerDeref` now returns `null` in the case that the pointer
deref cannot happen at compile-time. This is true for global
variables, for example. Another example is if a comptime known
pointer has a hard coded address value.
* Binary arithmetic sets the requireRuntimeBlock source location to the
lhs_src or rhs_src as appropriate instead of on the operator node.
* Fix LLVM codegen for slice_elem_val which had the wrong logic for
when the operand was not a pointer.
As noted in the comment in the implementation of deleteUnusedDecl, a
future improvement will be to rework the frontend/linker interface to
remove the frontend's responsibility of calling allocateDeclIndexes.
I discovered some issues with the plan9 linker backend that are related
to this, and worked around them for now.
2021-07-29 19:30:37 -07:00
|
|
|
|
|
|
|
|
|
2023-05-28 02:41:22 -04:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2020-09-13 19:17:58 -07:00
|
|
|
|
|
|
|
|
|
2023-01-30 18:22:50 +01:00
|
|
|
|
2022-11-27 01:07:35 -07:00
|
|
|
|
2020-09-13 19:17:58 -07:00
|
|
|
|
2022-11-27 01:07:35 -07:00
|
|
|
|
2021-06-03 15:39:26 -05:00
|
|
|
|
2020-09-13 19:17:58 -07:00
|
|
|
|
2022-11-27 01:07:35 -07:00
|
|
|
|
2020-09-13 19:17:58 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2022-04-19 21:51:08 -07:00
|
|
|
|
2020-09-13 19:17:58 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2022-11-27 01:07:35 -07:00
|
|
|
|
2020-09-13 19:17:58 -07:00
|
|
|
|
2021-06-03 15:39:26 -05:00
|
|
|
|
2020-09-13 19:17:58 -07:00
|
|
|
|
|
|
|
|
|
stage2: *WIP*: rework ZIR memory layout; overhaul source locations
The memory layout for ZIR instructions is completely reworked. See
zir.zig for those changes. Some new types:
* `zir.Code`: a "finished" set of ZIR instructions. Instead of allocating
each instruction independently, there is now a Tag and 8 bytes of
data available for all ZIR instructions. Small instructions fit
within these 8 bytes; larger ones use 4 bytes for an index into
`extra`. There is also `string_bytes` so that we can have 4 byte
references to strings. `zir.Inst.Tag` describes how to interpret
those 8 bytes of data.
- This is shared by all `Block` scopes.
* `Module.WipZirCode`: represents an in-progress `zir.Code`. In this
structure, the arrays are mutable, and get resized as we add/delete
things. There is extra state to keep track of things. This struct is
stored on the stack. Once it is finished, it produces an immutable
`zir.Code`, which will remain on the heap for the duration of a
function's existence.
- This is shared by all `GenZir` scopes.
* `Sema`: represents in-progress semantic analysis of a `zir.Code`.
This data is stored on the stack and is shared among all `Block`
scopes. It is now the main "self" argument to everything in the file
that was previously named `zir_sema.zig`.
Additionally, I moved some logic that was in `Module` into here.
`Module.Fn` now stores its parameter names inside the `zir.Code`,
instead of inside ZIR instructions. When the TZIR memory layout
reworking time comes, codegen will be able to reference this data
directly instead of duplicating it.
astgen.zig is (so far) almost entirely untouched, but nearly all of it
will need to be reworked to adhere to this new memory layout structure.
I have no benchmarks to report yet, as I am still working through
compile errors and fixing various things that I broke in this branch.
Overhaul of Source Locations:
Previously we used `usize` everywhere to mean byte offset, but sometimes
also mean other stuff. This was error prone and also made us do
unnecessary work, and store unnecessary bytes in memory.
Now there are more types involved into source locations, and more ways
to describe a source location.
* AllErrors.Message: embrace the assumption that files always have less
than 2 << 32 bytes.
* SrcLoc gets more complicated, to model more complicated source
locations.
* Introduce LazySrcLoc, which can model interesting source locations
with very little stored state. Useful for avoiding doing unnecessary
work when no compile errors occur.
Also, previously, we had `src: usize` on every ZIR instruction. This is
no longer the case. Each instruction now determines whether it even cares
about source location, and if so, how that source location is stored.
This requires more careful work inside `Sema`, but it results in fewer
bytes stored on the heap, without compromising accuracy and power of
compile error messages.
Miscellaneous:
* std.zig: string literals have more helpful result values for
reporting errors. There is now a lower level API and a higher level
API.
- side note: I noticed that the string literal logic needs some love.
There is some unnecessarily hacky code there.
* cut & pasted some TZIR logic that was in zir.zig to ir.zig. This
probably broke stuff and needs to get fixed.
* Removed type/Enum.zig, type/Union.zig, and type/Struct.zig. I don't
think this quite how this code will be organized. Need some more
careful planning about how to implement structs, unions, enums. They
need to be independent Decls, just like a top level function.
2021-03-15 23:38:38 -07:00
|
|
|
|
2023-06-12 01:44:12 -04:00
|
|
|
|
2020-10-07 20:32:02 +02:00
|
|
|
|
stage2: *WIP*: rework ZIR memory layout; overhaul source locations
The memory layout for ZIR instructions is completely reworked. See
zir.zig for those changes. Some new types:
* `zir.Code`: a "finished" set of ZIR instructions. Instead of allocating
each instruction independently, there is now a Tag and 8 bytes of
data available for all ZIR instructions. Small instructions fit
within these 8 bytes; larger ones use 4 bytes for an index into
`extra`. There is also `string_bytes` so that we can have 4 byte
references to strings. `zir.Inst.Tag` describes how to interpret
those 8 bytes of data.
- This is shared by all `Block` scopes.
* `Module.WipZirCode`: represents an in-progress `zir.Code`. In this
structure, the arrays are mutable, and get resized as we add/delete
things. There is extra state to keep track of things. This struct is
stored on the stack. Once it is finished, it produces an immutable
`zir.Code`, which will remain on the heap for the duration of a
function's existence.
- This is shared by all `GenZir` scopes.
* `Sema`: represents in-progress semantic analysis of a `zir.Code`.
This data is stored on the stack and is shared among all `Block`
scopes. It is now the main "self" argument to everything in the file
that was previously named `zir_sema.zig`.
Additionally, I moved some logic that was in `Module` into here.
`Module.Fn` now stores its parameter names inside the `zir.Code`,
instead of inside ZIR instructions. When the TZIR memory layout
reworking time comes, codegen will be able to reference this data
directly instead of duplicating it.
astgen.zig is (so far) almost entirely untouched, but nearly all of it
will need to be reworked to adhere to this new memory layout structure.
I have no benchmarks to report yet, as I am still working through
compile errors and fixing various things that I broke in this branch.
Overhaul of Source Locations:
Previously we used `usize` everywhere to mean byte offset, but sometimes
also mean other stuff. This was error prone and also made us do
unnecessary work, and store unnecessary bytes in memory.
Now there are more types involved into source locations, and more ways
to describe a source location.
* AllErrors.Message: embrace the assumption that files always have less
than 2 << 32 bytes.
* SrcLoc gets more complicated, to model more complicated source
locations.
* Introduce LazySrcLoc, which can model interesting source locations
with very little stored state. Useful for avoiding doing unnecessary
work when no compile errors occur.
Also, previously, we had `src: usize` on every ZIR instruction. This is
no longer the case. Each instruction now determines whether it even cares
about source location, and if so, how that source location is stored.
This requires more careful work inside `Sema`, but it results in fewer
bytes stored on the heap, without compromising accuracy and power of
compile error messages.
Miscellaneous:
* std.zig: string literals have more helpful result values for
reporting errors. There is now a lower level API and a higher level
API.
- side note: I noticed that the string literal logic needs some love.
There is some unnecessarily hacky code there.
* cut & pasted some TZIR logic that was in zir.zig to ir.zig. This
probably broke stuff and needs to get fixed.
* Removed type/Enum.zig, type/Union.zig, and type/Struct.zig. I don't
think this quite how this code will be organized. Need some more
careful planning about how to implement structs, unions, enums. They
need to be independent Decls, just like a top level function.
2021-03-15 23:38:38 -07:00
|
|
|
|
2023-06-12 01:44:12 -04:00
|
|
|
|
2020-09-13 19:17:58 -07:00
|
|
|
|
2022-02-21 21:43:38 +01:00
|
|
|
|
2023-02-01 18:55:35 +01:00
|
|
|
|
2022-02-21 21:43:38 +01:00
|
|
|
|
2022-08-29 08:50:39 +02:00
|
|
|
|
2023-06-12 01:44:12 -04:00
|
|
|
|
2022-08-29 08:50:39 +02:00
|
|
|
|
2021-06-03 15:39:26 -05:00
|
|
|
|
|
|
|
|
|
2020-09-13 19:17:58 -07:00
|
|
|
|
stage2: *WIP*: rework ZIR memory layout; overhaul source locations
The memory layout for ZIR instructions is completely reworked. See
zir.zig for those changes. Some new types:
* `zir.Code`: a "finished" set of ZIR instructions. Instead of allocating
each instruction independently, there is now a Tag and 8 bytes of
data available for all ZIR instructions. Small instructions fit
within these 8 bytes; larger ones use 4 bytes for an index into
`extra`. There is also `string_bytes` so that we can have 4 byte
references to strings. `zir.Inst.Tag` describes how to interpret
those 8 bytes of data.
- This is shared by all `Block` scopes.
* `Module.WipZirCode`: represents an in-progress `zir.Code`. In this
structure, the arrays are mutable, and get resized as we add/delete
things. There is extra state to keep track of things. This struct is
stored on the stack. Once it is finished, it produces an immutable
`zir.Code`, which will remain on the heap for the duration of a
function's existence.
- This is shared by all `GenZir` scopes.
* `Sema`: represents in-progress semantic analysis of a `zir.Code`.
This data is stored on the stack and is shared among all `Block`
scopes. It is now the main "self" argument to everything in the file
that was previously named `zir_sema.zig`.
Additionally, I moved some logic that was in `Module` into here.
`Module.Fn` now stores its parameter names inside the `zir.Code`,
instead of inside ZIR instructions. When the TZIR memory layout
reworking time comes, codegen will be able to reference this data
directly instead of duplicating it.
astgen.zig is (so far) almost entirely untouched, but nearly all of it
will need to be reworked to adhere to this new memory layout structure.
I have no benchmarks to report yet, as I am still working through
compile errors and fixing various things that I broke in this branch.
Overhaul of Source Locations:
Previously we used `usize` everywhere to mean byte offset, but sometimes
also mean other stuff. This was error prone and also made us do
unnecessary work, and store unnecessary bytes in memory.
Now there are more types involved into source locations, and more ways
to describe a source location.
* AllErrors.Message: embrace the assumption that files always have less
than 2 << 32 bytes.
* SrcLoc gets more complicated, to model more complicated source
locations.
* Introduce LazySrcLoc, which can model interesting source locations
with very little stored state. Useful for avoiding doing unnecessary
work when no compile errors occur.
Also, previously, we had `src: usize` on every ZIR instruction. This is
no longer the case. Each instruction now determines whether it even cares
about source location, and if so, how that source location is stored.
This requires more careful work inside `Sema`, but it results in fewer
bytes stored on the heap, without compromising accuracy and power of
compile error messages.
Miscellaneous:
* std.zig: string literals have more helpful result values for
reporting errors. There is now a lower level API and a higher level
API.
- side note: I noticed that the string literal logic needs some love.
There is some unnecessarily hacky code there.
* cut & pasted some TZIR logic that was in zir.zig to ir.zig. This
probably broke stuff and needs to get fixed.
* Removed type/Enum.zig, type/Union.zig, and type/Struct.zig. I don't
think this quite how this code will be organized. Need some more
careful planning about how to implement structs, unions, enums. They
need to be independent Decls, just like a top level function.
2021-03-15 23:38:38 -07:00
|
|
|
|
2020-09-13 19:17:58 -07:00
|
|
|
|
2022-11-27 01:07:35 -07:00
|
|
|
|
2020-09-13 19:17:58 -07:00
|
|
|
|
|
|
|
|
|
rework generic function calls
Abridged summary:
* Move `Module.Fn` into `InternPool`.
* Delete a lot of confusing and problematic `Sema` logic related to
generic function calls.
This commit removes `Module.Fn` and replaces it with two new
`InternPool.Tag` values:
* `func_decl` - corresponding to a function declared in the source
code. This one contains line/column numbers, zir_body_inst, etc.
* `func_instance` - one for each monomorphization of a generic
function. Contains a reference to the `func_decl` from whence the
instantiation came, along with the `comptime` parameter values (or
types in the case of `anytype`)
Since `InternPool` provides deduplication on these values, these fields
are now deleted from `Module`:
* `monomorphed_func_keys`
* `monomorphed_funcs`
* `align_stack_fns`
Instead of these, Sema logic for generic function instantiation now
unconditionally evaluates the function prototype expression for every
generic callsite. This is technically required in order for type
coercions to work. The previous code had some dubious, probably wrong
hacks to make things work, such as `hashUncoerced`. I'm not 100% sure
how we were able to eliminate that function and still pass all the
behavior tests, but I'm pretty sure things were still broken without
doing type coercion for every generic function call argument.
After the function prototype is evaluated, it produces a deduplicated
`func_instance` `InternPool.Index` which can then be used for the
generic function call.
Some other nice things made by this simplification are the removal of
`comptime_args_fn_inst` and `preallocated_new_func` from `Sema`, and the
messy logic associated with them.
I have not yet been able to measure the perf of this against master
branch. On one hand, it reduces memory usage and pointer chasing of the
most heavily used `InternPool` Tag - function bodies - but on the other
hand, it does evaluate function prototype expressions more than before.
We will soon find out.
2023-07-03 22:09:30 -07:00
|
|
|
|
2020-09-13 19:17:58 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-07-11 16:32:11 -07:00
|
|
|
|
rework generic function calls
Abridged summary:
* Move `Module.Fn` into `InternPool`.
* Delete a lot of confusing and problematic `Sema` logic related to
generic function calls.
This commit removes `Module.Fn` and replaces it with two new
`InternPool.Tag` values:
* `func_decl` - corresponding to a function declared in the source
code. This one contains line/column numbers, zir_body_inst, etc.
* `func_instance` - one for each monomorphization of a generic
function. Contains a reference to the `func_decl` from whence the
instantiation came, along with the `comptime` parameter values (or
types in the case of `anytype`)
Since `InternPool` provides deduplication on these values, these fields
are now deleted from `Module`:
* `monomorphed_func_keys`
* `monomorphed_funcs`
* `align_stack_fns`
Instead of these, Sema logic for generic function instantiation now
unconditionally evaluates the function prototype expression for every
generic callsite. This is technically required in order for type
coercions to work. The previous code had some dubious, probably wrong
hacks to make things work, such as `hashUncoerced`. I'm not 100% sure
how we were able to eliminate that function and still pass all the
behavior tests, but I'm pretty sure things were still broken without
doing type coercion for every generic function call argument.
After the function prototype is evaluated, it produces a deduplicated
`func_instance` `InternPool.Index` which can then be used for the
generic function call.
Some other nice things made by this simplification are the removal of
`comptime_args_fn_inst` and `preallocated_new_func` from `Sema`, and the
messy logic associated with them.
I have not yet been able to measure the perf of this against master
branch. On one hand, it reduces memory usage and pointer chasing of the
most heavily used `InternPool` Tag - function bodies - but on the other
hand, it does evaluate function prototype expressions more than before.
We will soon find out.
2023-07-03 22:09:30 -07:00
|
|
|
|
|
|
|
|
|
2022-04-19 21:51:08 -07:00
|
|
|
|
|
|
|
|
|
2021-07-11 16:32:11 -07:00
|
|
|
|
2023-06-02 01:55:16 +01:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
rework generic function calls
Abridged summary:
* Move `Module.Fn` into `InternPool`.
* Delete a lot of confusing and problematic `Sema` logic related to
generic function calls.
This commit removes `Module.Fn` and replaces it with two new
`InternPool.Tag` values:
* `func_decl` - corresponding to a function declared in the source
code. This one contains line/column numbers, zir_body_inst, etc.
* `func_instance` - one for each monomorphization of a generic
function. Contains a reference to the `func_decl` from whence the
instantiation came, along with the `comptime` parameter values (or
types in the case of `anytype`)
Since `InternPool` provides deduplication on these values, these fields
are now deleted from `Module`:
* `monomorphed_func_keys`
* `monomorphed_funcs`
* `align_stack_fns`
Instead of these, Sema logic for generic function instantiation now
unconditionally evaluates the function prototype expression for every
generic callsite. This is technically required in order for type
coercions to work. The previous code had some dubious, probably wrong
hacks to make things work, such as `hashUncoerced`. I'm not 100% sure
how we were able to eliminate that function and still pass all the
behavior tests, but I'm pretty sure things were still broken without
doing type coercion for every generic function call argument.
After the function prototype is evaluated, it produces a deduplicated
`func_instance` `InternPool.Index` which can then be used for the
generic function call.
Some other nice things made by this simplification are the removal of
`comptime_args_fn_inst` and `preallocated_new_func` from `Sema`, and the
messy logic associated with them.
I have not yet been able to measure the perf of this against master
branch. On one hand, it reduces memory usage and pointer chasing of the
most heavily used `InternPool` Tag - function bodies - but on the other
hand, it does evaluate function prototype expressions more than before.
We will soon find out.
2023-07-03 22:09:30 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-05-15 20:09:54 -07:00
|
|
|
|
rework generic function calls
Abridged summary:
* Move `Module.Fn` into `InternPool`.
* Delete a lot of confusing and problematic `Sema` logic related to
generic function calls.
This commit removes `Module.Fn` and replaces it with two new
`InternPool.Tag` values:
* `func_decl` - corresponding to a function declared in the source
code. This one contains line/column numbers, zir_body_inst, etc.
* `func_instance` - one for each monomorphization of a generic
function. Contains a reference to the `func_decl` from whence the
instantiation came, along with the `comptime` parameter values (or
types in the case of `anytype`)
Since `InternPool` provides deduplication on these values, these fields
are now deleted from `Module`:
* `monomorphed_func_keys`
* `monomorphed_funcs`
* `align_stack_fns`
Instead of these, Sema logic for generic function instantiation now
unconditionally evaluates the function prototype expression for every
generic callsite. This is technically required in order for type
coercions to work. The previous code had some dubious, probably wrong
hacks to make things work, such as `hashUncoerced`. I'm not 100% sure
how we were able to eliminate that function and still pass all the
behavior tests, but I'm pretty sure things were still broken without
doing type coercion for every generic function call argument.
After the function prototype is evaluated, it produces a deduplicated
`func_instance` `InternPool.Index` which can then be used for the
generic function call.
Some other nice things made by this simplification are the removal of
`comptime_args_fn_inst` and `preallocated_new_func` from `Sema`, and the
messy logic associated with them.
I have not yet been able to measure the perf of this against master
branch. On one hand, it reduces memory usage and pointer chasing of the
most heavily used `InternPool` Tag - function bodies - but on the other
hand, it does evaluate function prototype expressions more than before.
We will soon find out.
2023-07-03 22:09:30 -07:00
|
|
|
|
2023-05-15 20:09:54 -07:00
|
|
|
|
2021-04-30 23:11:20 -07:00
|
|
|
|
|
|
|
|
|
2021-07-11 16:32:11 -07:00
|
|
|
|
2021-09-21 15:08:32 -07:00
|
|
|
|
2023-05-08 16:52:59 -07:00
|
|
|
|
2021-04-30 23:11:20 -07:00
|
|
|
|
2022-04-19 21:51:08 -07:00
|
|
|
|
rework generic function calls
Abridged summary:
* Move `Module.Fn` into `InternPool`.
* Delete a lot of confusing and problematic `Sema` logic related to
generic function calls.
This commit removes `Module.Fn` and replaces it with two new
`InternPool.Tag` values:
* `func_decl` - corresponding to a function declared in the source
code. This one contains line/column numbers, zir_body_inst, etc.
* `func_instance` - one for each monomorphization of a generic
function. Contains a reference to the `func_decl` from whence the
instantiation came, along with the `comptime` parameter values (or
types in the case of `anytype`)
Since `InternPool` provides deduplication on these values, these fields
are now deleted from `Module`:
* `monomorphed_func_keys`
* `monomorphed_funcs`
* `align_stack_fns`
Instead of these, Sema logic for generic function instantiation now
unconditionally evaluates the function prototype expression for every
generic callsite. This is technically required in order for type
coercions to work. The previous code had some dubious, probably wrong
hacks to make things work, such as `hashUncoerced`. I'm not 100% sure
how we were able to eliminate that function and still pass all the
behavior tests, but I'm pretty sure things were still broken without
doing type coercion for every generic function call argument.
After the function prototype is evaluated, it produces a deduplicated
`func_instance` `InternPool.Index` which can then be used for the
generic function call.
Some other nice things made by this simplification are the removal of
`comptime_args_fn_inst` and `preallocated_new_func` from `Sema`, and the
messy logic associated with them.
I have not yet been able to measure the perf of this against master
branch. On one hand, it reduces memory usage and pointer chasing of the
most heavily used `InternPool` Tag - function bodies - but on the other
hand, it does evaluate function prototype expressions more than before.
We will soon find out.
2023-07-03 22:09:30 -07:00
|
|
|
|
2023-07-31 01:50:54 -04:00
|
|
|
|
rework generic function calls
Abridged summary:
* Move `Module.Fn` into `InternPool`.
* Delete a lot of confusing and problematic `Sema` logic related to
generic function calls.
This commit removes `Module.Fn` and replaces it with two new
`InternPool.Tag` values:
* `func_decl` - corresponding to a function declared in the source
code. This one contains line/column numbers, zir_body_inst, etc.
* `func_instance` - one for each monomorphization of a generic
function. Contains a reference to the `func_decl` from whence the
instantiation came, along with the `comptime` parameter values (or
types in the case of `anytype`)
Since `InternPool` provides deduplication on these values, these fields
are now deleted from `Module`:
* `monomorphed_func_keys`
* `monomorphed_funcs`
* `align_stack_fns`
Instead of these, Sema logic for generic function instantiation now
unconditionally evaluates the function prototype expression for every
generic callsite. This is technically required in order for type
coercions to work. The previous code had some dubious, probably wrong
hacks to make things work, such as `hashUncoerced`. I'm not 100% sure
how we were able to eliminate that function and still pass all the
behavior tests, but I'm pretty sure things were still broken without
doing type coercion for every generic function call argument.
After the function prototype is evaluated, it produces a deduplicated
`func_instance` `InternPool.Index` which can then be used for the
generic function call.
Some other nice things made by this simplification are the removal of
`comptime_args_fn_inst` and `preallocated_new_func` from `Sema`, and the
messy logic associated with them.
I have not yet been able to measure the perf of this against master
branch. On one hand, it reduces memory usage and pointer chasing of the
most heavily used `InternPool` Tag - function bodies - but on the other
hand, it does evaluate function prototype expressions more than before.
We will soon find out.
2023-07-03 22:09:30 -07:00
|
|
|
|
2023-07-08 23:39:37 -07:00
|
|
|
|
rework generic function calls
Abridged summary:
* Move `Module.Fn` into `InternPool`.
* Delete a lot of confusing and problematic `Sema` logic related to
generic function calls.
This commit removes `Module.Fn` and replaces it with two new
`InternPool.Tag` values:
* `func_decl` - corresponding to a function declared in the source
code. This one contains line/column numbers, zir_body_inst, etc.
* `func_instance` - one for each monomorphization of a generic
function. Contains a reference to the `func_decl` from whence the
instantiation came, along with the `comptime` parameter values (or
types in the case of `anytype`)
Since `InternPool` provides deduplication on these values, these fields
are now deleted from `Module`:
* `monomorphed_func_keys`
* `monomorphed_funcs`
* `align_stack_fns`
Instead of these, Sema logic for generic function instantiation now
unconditionally evaluates the function prototype expression for every
generic callsite. This is technically required in order for type
coercions to work. The previous code had some dubious, probably wrong
hacks to make things work, such as `hashUncoerced`. I'm not 100% sure
how we were able to eliminate that function and still pass all the
behavior tests, but I'm pretty sure things were still broken without
doing type coercion for every generic function call argument.
After the function prototype is evaluated, it produces a deduplicated
`func_instance` `InternPool.Index` which can then be used for the
generic function call.
Some other nice things made by this simplification are the removal of
`comptime_args_fn_inst` and `preallocated_new_func` from `Sema`, and the
messy logic associated with them.
I have not yet been able to measure the perf of this against master
branch. On one hand, it reduces memory usage and pointer chasing of the
most heavily used `InternPool` Tag - function bodies - but on the other
hand, it does evaluate function prototype expressions more than before.
We will soon find out.
2023-07-03 22:09:30 -07:00
|
|
|
|
|
|
|
|
|
2023-06-02 01:55:16 +01:00
|
|
|
|
2021-04-30 23:11:20 -07:00
|
|
|
|
2021-05-17 17:39:52 -07:00
|
|
|
|
2020-09-13 19:17:58 -07:00
|
|
|
|
2023-07-08 23:39:37 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2022-04-23 14:48:16 +03:00
|
|
|
|
rework generic function calls
Abridged summary:
* Move `Module.Fn` into `InternPool`.
* Delete a lot of confusing and problematic `Sema` logic related to
generic function calls.
This commit removes `Module.Fn` and replaces it with two new
`InternPool.Tag` values:
* `func_decl` - corresponding to a function declared in the source
code. This one contains line/column numbers, zir_body_inst, etc.
* `func_instance` - one for each monomorphization of a generic
function. Contains a reference to the `func_decl` from whence the
instantiation came, along with the `comptime` parameter values (or
types in the case of `anytype`)
Since `InternPool` provides deduplication on these values, these fields
are now deleted from `Module`:
* `monomorphed_func_keys`
* `monomorphed_funcs`
* `align_stack_fns`
Instead of these, Sema logic for generic function instantiation now
unconditionally evaluates the function prototype expression for every
generic callsite. This is technically required in order for type
coercions to work. The previous code had some dubious, probably wrong
hacks to make things work, such as `hashUncoerced`. I'm not 100% sure
how we were able to eliminate that function and still pass all the
behavior tests, but I'm pretty sure things were still broken without
doing type coercion for every generic function call argument.
After the function prototype is evaluated, it produces a deduplicated
`func_instance` `InternPool.Index` which can then be used for the
generic function call.
Some other nice things made by this simplification are the removal of
`comptime_args_fn_inst` and `preallocated_new_func` from `Sema`, and the
messy logic associated with them.
I have not yet been able to measure the perf of this against master
branch. On one hand, it reduces memory usage and pointer chasing of the
most heavily used `InternPool` Tag - function bodies - but on the other
hand, it does evaluate function prototype expressions more than before.
We will soon find out.
2023-07-03 22:09:30 -07:00
|
|
|
|
2022-04-22 21:30:54 +03:00
|
|
|
|
2021-07-11 16:32:11 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-10-01 17:45:13 -05:00
|
|
|
|
2021-04-30 23:11:20 -07:00
|
|
|
|
|
|
|
|
|
2022-04-19 21:51:08 -07:00
|
|
|
|
2021-10-01 12:05:12 -05:00
|
|
|
|
2023-09-14 19:59:58 -07:00
|
|
|
|
2021-04-30 23:11:20 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-07-11 16:32:11 -07:00
|
|
|
|
2021-03-22 23:46:51 -07:00
|
|
|
|
rework generic function calls
Abridged summary:
* Move `Module.Fn` into `InternPool`.
* Delete a lot of confusing and problematic `Sema` logic related to
generic function calls.
This commit removes `Module.Fn` and replaces it with two new
`InternPool.Tag` values:
* `func_decl` - corresponding to a function declared in the source
code. This one contains line/column numbers, zir_body_inst, etc.
* `func_instance` - one for each monomorphization of a generic
function. Contains a reference to the `func_decl` from whence the
instantiation came, along with the `comptime` parameter values (or
types in the case of `anytype`)
Since `InternPool` provides deduplication on these values, these fields
are now deleted from `Module`:
* `monomorphed_func_keys`
* `monomorphed_funcs`
* `align_stack_fns`
Instead of these, Sema logic for generic function instantiation now
unconditionally evaluates the function prototype expression for every
generic callsite. This is technically required in order for type
coercions to work. The previous code had some dubious, probably wrong
hacks to make things work, such as `hashUncoerced`. I'm not 100% sure
how we were able to eliminate that function and still pass all the
behavior tests, but I'm pretty sure things were still broken without
doing type coercion for every generic function call argument.
After the function prototype is evaluated, it produces a deduplicated
`func_instance` `InternPool.Index` which can then be used for the
generic function call.
Some other nice things made by this simplification are the removal of
`comptime_args_fn_inst` and `preallocated_new_func` from `Sema`, and the
messy logic associated with them.
I have not yet been able to measure the perf of this against master
branch. On one hand, it reduces memory usage and pointer chasing of the
most heavily used `InternPool` Tag - function bodies - but on the other
hand, it does evaluate function prototype expressions more than before.
We will soon find out.
2023-07-03 22:09:30 -07:00
|
|
|
|
2021-08-03 17:29:59 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-08-03 22:34:22 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
rework generic function calls
Abridged summary:
* Move `Module.Fn` into `InternPool`.
* Delete a lot of confusing and problematic `Sema` logic related to
generic function calls.
This commit removes `Module.Fn` and replaces it with two new
`InternPool.Tag` values:
* `func_decl` - corresponding to a function declared in the source
code. This one contains line/column numbers, zir_body_inst, etc.
* `func_instance` - one for each monomorphization of a generic
function. Contains a reference to the `func_decl` from whence the
instantiation came, along with the `comptime` parameter values (or
types in the case of `anytype`)
Since `InternPool` provides deduplication on these values, these fields
are now deleted from `Module`:
* `monomorphed_func_keys`
* `monomorphed_funcs`
* `align_stack_fns`
Instead of these, Sema logic for generic function instantiation now
unconditionally evaluates the function prototype expression for every
generic callsite. This is technically required in order for type
coercions to work. The previous code had some dubious, probably wrong
hacks to make things work, such as `hashUncoerced`. I'm not 100% sure
how we were able to eliminate that function and still pass all the
behavior tests, but I'm pretty sure things were still broken without
doing type coercion for every generic function call argument.
After the function prototype is evaluated, it produces a deduplicated
`func_instance` `InternPool.Index` which can then be used for the
generic function call.
Some other nice things made by this simplification are the removal of
`comptime_args_fn_inst` and `preallocated_new_func` from `Sema`, and the
messy logic associated with them.
I have not yet been able to measure the perf of this against master
branch. On one hand, it reduces memory usage and pointer chasing of the
most heavily used `InternPool` Tag - function bodies - but on the other
hand, it does evaluate function prototype expressions more than before.
We will soon find out.
2023-07-03 22:09:30 -07:00
|
|
|
|
2021-11-01 00:56:55 -07:00
|
|
|
|
rework generic function calls
Abridged summary:
* Move `Module.Fn` into `InternPool`.
* Delete a lot of confusing and problematic `Sema` logic related to
generic function calls.
This commit removes `Module.Fn` and replaces it with two new
`InternPool.Tag` values:
* `func_decl` - corresponding to a function declared in the source
code. This one contains line/column numbers, zir_body_inst, etc.
* `func_instance` - one for each monomorphization of a generic
function. Contains a reference to the `func_decl` from whence the
instantiation came, along with the `comptime` parameter values (or
types in the case of `anytype`)
Since `InternPool` provides deduplication on these values, these fields
are now deleted from `Module`:
* `monomorphed_func_keys`
* `monomorphed_funcs`
* `align_stack_fns`
Instead of these, Sema logic for generic function instantiation now
unconditionally evaluates the function prototype expression for every
generic callsite. This is technically required in order for type
coercions to work. The previous code had some dubious, probably wrong
hacks to make things work, such as `hashUncoerced`. I'm not 100% sure
how we were able to eliminate that function and still pass all the
behavior tests, but I'm pretty sure things were still broken without
doing type coercion for every generic function call argument.
After the function prototype is evaluated, it produces a deduplicated
`func_instance` `InternPool.Index` which can then be used for the
generic function call.
Some other nice things made by this simplification are the removal of
`comptime_args_fn_inst` and `preallocated_new_func` from `Sema`, and the
messy logic associated with them.
I have not yet been able to measure the perf of this against master
branch. On one hand, it reduces memory usage and pointer chasing of the
most heavily used `InternPool` Tag - function bodies - but on the other
hand, it does evaluate function prototype expressions more than before.
We will soon find out.
2023-07-03 22:09:30 -07:00
|
|
|
|
2022-11-25 20:12:25 +01:00
|
|
|
|
2021-08-03 17:29:59 -07:00
|
|
|
|
rework generic function calls
Abridged summary:
* Move `Module.Fn` into `InternPool`.
* Delete a lot of confusing and problematic `Sema` logic related to
generic function calls.
This commit removes `Module.Fn` and replaces it with two new
`InternPool.Tag` values:
* `func_decl` - corresponding to a function declared in the source
code. This one contains line/column numbers, zir_body_inst, etc.
* `func_instance` - one for each monomorphization of a generic
function. Contains a reference to the `func_decl` from whence the
instantiation came, along with the `comptime` parameter values (or
types in the case of `anytype`)
Since `InternPool` provides deduplication on these values, these fields
are now deleted from `Module`:
* `monomorphed_func_keys`
* `monomorphed_funcs`
* `align_stack_fns`
Instead of these, Sema logic for generic function instantiation now
unconditionally evaluates the function prototype expression for every
generic callsite. This is technically required in order for type
coercions to work. The previous code had some dubious, probably wrong
hacks to make things work, such as `hashUncoerced`. I'm not 100% sure
how we were able to eliminate that function and still pass all the
behavior tests, but I'm pretty sure things were still broken without
doing type coercion for every generic function call argument.
After the function prototype is evaluated, it produces a deduplicated
`func_instance` `InternPool.Index` which can then be used for the
generic function call.
Some other nice things made by this simplification are the removal of
`comptime_args_fn_inst` and `preallocated_new_func` from `Sema`, and the
messy logic associated with them.
I have not yet been able to measure the perf of this against master
branch. On one hand, it reduces memory usage and pointer chasing of the
most heavily used `InternPool` Tag - function bodies - but on the other
hand, it does evaluate function prototype expressions more than before.
We will soon find out.
2023-07-03 22:09:30 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2022-11-12 18:23:40 +02:00
|
|
|
|
rework generic function calls
Abridged summary:
* Move `Module.Fn` into `InternPool`.
* Delete a lot of confusing and problematic `Sema` logic related to
generic function calls.
This commit removes `Module.Fn` and replaces it with two new
`InternPool.Tag` values:
* `func_decl` - corresponding to a function declared in the source
code. This one contains line/column numbers, zir_body_inst, etc.
* `func_instance` - one for each monomorphization of a generic
function. Contains a reference to the `func_decl` from whence the
instantiation came, along with the `comptime` parameter values (or
types in the case of `anytype`)
Since `InternPool` provides deduplication on these values, these fields
are now deleted from `Module`:
* `monomorphed_func_keys`
* `monomorphed_funcs`
* `align_stack_fns`
Instead of these, Sema logic for generic function instantiation now
unconditionally evaluates the function prototype expression for every
generic callsite. This is technically required in order for type
coercions to work. The previous code had some dubious, probably wrong
hacks to make things work, such as `hashUncoerced`. I'm not 100% sure
how we were able to eliminate that function and still pass all the
behavior tests, but I'm pretty sure things were still broken without
doing type coercion for every generic function call argument.
After the function prototype is evaluated, it produces a deduplicated
`func_instance` `InternPool.Index` which can then be used for the
generic function call.
Some other nice things made by this simplification are the removal of
`comptime_args_fn_inst` and `preallocated_new_func` from `Sema`, and the
messy logic associated with them.
I have not yet been able to measure the perf of this against master
branch. On one hand, it reduces memory usage and pointer chasing of the
most heavily used `InternPool` Tag - function bodies - but on the other
hand, it does evaluate function prototype expressions more than before.
We will soon find out.
2023-07-03 22:09:30 -07:00
|
|
|
|
2022-04-14 00:36:54 -07:00
|
|
|
|
rework generic function calls
Abridged summary:
* Move `Module.Fn` into `InternPool`.
* Delete a lot of confusing and problematic `Sema` logic related to
generic function calls.
This commit removes `Module.Fn` and replaces it with two new
`InternPool.Tag` values:
* `func_decl` - corresponding to a function declared in the source
code. This one contains line/column numbers, zir_body_inst, etc.
* `func_instance` - one for each monomorphization of a generic
function. Contains a reference to the `func_decl` from whence the
instantiation came, along with the `comptime` parameter values (or
types in the case of `anytype`)
Since `InternPool` provides deduplication on these values, these fields
are now deleted from `Module`:
* `monomorphed_func_keys`
* `monomorphed_funcs`
* `align_stack_fns`
Instead of these, Sema logic for generic function instantiation now
unconditionally evaluates the function prototype expression for every
generic callsite. This is technically required in order for type
coercions to work. The previous code had some dubious, probably wrong
hacks to make things work, such as `hashUncoerced`. I'm not 100% sure
how we were able to eliminate that function and still pass all the
behavior tests, but I'm pretty sure things were still broken without
doing type coercion for every generic function call argument.
After the function prototype is evaluated, it produces a deduplicated
`func_instance` `InternPool.Index` which can then be used for the
generic function call.
Some other nice things made by this simplification are the removal of
`comptime_args_fn_inst` and `preallocated_new_func` from `Sema`, and the
messy logic associated with them.
I have not yet been able to measure the perf of this against master
branch. On one hand, it reduces memory usage and pointer chasing of the
most heavily used `InternPool` Tag - function bodies - but on the other
hand, it does evaluate function prototype expressions more than before.
We will soon find out.
2023-07-03 22:09:30 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2022-04-14 00:36:54 -07:00
|
|
|
|
rework generic function calls
Abridged summary:
* Move `Module.Fn` into `InternPool`.
* Delete a lot of confusing and problematic `Sema` logic related to
generic function calls.
This commit removes `Module.Fn` and replaces it with two new
`InternPool.Tag` values:
* `func_decl` - corresponding to a function declared in the source
code. This one contains line/column numbers, zir_body_inst, etc.
* `func_instance` - one for each monomorphization of a generic
function. Contains a reference to the `func_decl` from whence the
instantiation came, along with the `comptime` parameter values (or
types in the case of `anytype`)
Since `InternPool` provides deduplication on these values, these fields
are now deleted from `Module`:
* `monomorphed_func_keys`
* `monomorphed_funcs`
* `align_stack_fns`
Instead of these, Sema logic for generic function instantiation now
unconditionally evaluates the function prototype expression for every
generic callsite. This is technically required in order for type
coercions to work. The previous code had some dubious, probably wrong
hacks to make things work, such as `hashUncoerced`. I'm not 100% sure
how we were able to eliminate that function and still pass all the
behavior tests, but I'm pretty sure things were still broken without
doing type coercion for every generic function call argument.
After the function prototype is evaluated, it produces a deduplicated
`func_instance` `InternPool.Index` which can then be used for the
generic function call.
Some other nice things made by this simplification are the removal of
`comptime_args_fn_inst` and `preallocated_new_func` from `Sema`, and the
messy logic associated with them.
I have not yet been able to measure the perf of this against master
branch. On one hand, it reduces memory usage and pointer chasing of the
most heavily used `InternPool` Tag - function bodies - but on the other
hand, it does evaluate function prototype expressions more than before.
We will soon find out.
2023-07-03 22:09:30 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2022-04-14 00:36:54 -07:00
|
|
|
|
rework generic function calls
Abridged summary:
* Move `Module.Fn` into `InternPool`.
* Delete a lot of confusing and problematic `Sema` logic related to
generic function calls.
This commit removes `Module.Fn` and replaces it with two new
`InternPool.Tag` values:
* `func_decl` - corresponding to a function declared in the source
code. This one contains line/column numbers, zir_body_inst, etc.
* `func_instance` - one for each monomorphization of a generic
function. Contains a reference to the `func_decl` from whence the
instantiation came, along with the `comptime` parameter values (or
types in the case of `anytype`)
Since `InternPool` provides deduplication on these values, these fields
are now deleted from `Module`:
* `monomorphed_func_keys`
* `monomorphed_funcs`
* `align_stack_fns`
Instead of these, Sema logic for generic function instantiation now
unconditionally evaluates the function prototype expression for every
generic callsite. This is technically required in order for type
coercions to work. The previous code had some dubious, probably wrong
hacks to make things work, such as `hashUncoerced`. I'm not 100% sure
how we were able to eliminate that function and still pass all the
behavior tests, but I'm pretty sure things were still broken without
doing type coercion for every generic function call argument.
After the function prototype is evaluated, it produces a deduplicated
`func_instance` `InternPool.Index` which can then be used for the
generic function call.
Some other nice things made by this simplification are the removal of
`comptime_args_fn_inst` and `preallocated_new_func` from `Sema`, and the
messy logic associated with them.
I have not yet been able to measure the perf of this against master
branch. On one hand, it reduces memory usage and pointer chasing of the
most heavily used `InternPool` Tag - function bodies - but on the other
hand, it does evaluate function prototype expressions more than before.
We will soon find out.
2023-07-03 22:09:30 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-12-27 17:56:33 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2022-01-17 15:21:58 -07:00
|
|
|
|
2021-12-27 17:56:33 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
rework generic function calls
Abridged summary:
* Move `Module.Fn` into `InternPool`.
* Delete a lot of confusing and problematic `Sema` logic related to
generic function calls.
This commit removes `Module.Fn` and replaces it with two new
`InternPool.Tag` values:
* `func_decl` - corresponding to a function declared in the source
code. This one contains line/column numbers, zir_body_inst, etc.
* `func_instance` - one for each monomorphization of a generic
function. Contains a reference to the `func_decl` from whence the
instantiation came, along with the `comptime` parameter values (or
types in the case of `anytype`)
Since `InternPool` provides deduplication on these values, these fields
are now deleted from `Module`:
* `monomorphed_func_keys`
* `monomorphed_funcs`
* `align_stack_fns`
Instead of these, Sema logic for generic function instantiation now
unconditionally evaluates the function prototype expression for every
generic callsite. This is technically required in order for type
coercions to work. The previous code had some dubious, probably wrong
hacks to make things work, such as `hashUncoerced`. I'm not 100% sure
how we were able to eliminate that function and still pass all the
behavior tests, but I'm pretty sure things were still broken without
doing type coercion for every generic function call argument.
After the function prototype is evaluated, it produces a deduplicated
`func_instance` `InternPool.Index` which can then be used for the
generic function call.
Some other nice things made by this simplification are the removal of
`comptime_args_fn_inst` and `preallocated_new_func` from `Sema`, and the
messy logic associated with them.
I have not yet been able to measure the perf of this against master
branch. On one hand, it reduces memory usage and pointer chasing of the
most heavily used `InternPool` Tag - function bodies - but on the other
hand, it does evaluate function prototype expressions more than before.
We will soon find out.
2023-07-03 22:09:30 -07:00
|
|
|
|
2021-12-27 17:56:33 -07:00
|
|
|
|
|
|
|
|
|
rework generic function calls
Abridged summary:
* Move `Module.Fn` into `InternPool`.
* Delete a lot of confusing and problematic `Sema` logic related to
generic function calls.
This commit removes `Module.Fn` and replaces it with two new
`InternPool.Tag` values:
* `func_decl` - corresponding to a function declared in the source
code. This one contains line/column numbers, zir_body_inst, etc.
* `func_instance` - one for each monomorphization of a generic
function. Contains a reference to the `func_decl` from whence the
instantiation came, along with the `comptime` parameter values (or
types in the case of `anytype`)
Since `InternPool` provides deduplication on these values, these fields
are now deleted from `Module`:
* `monomorphed_func_keys`
* `monomorphed_funcs`
* `align_stack_fns`
Instead of these, Sema logic for generic function instantiation now
unconditionally evaluates the function prototype expression for every
generic callsite. This is technically required in order for type
coercions to work. The previous code had some dubious, probably wrong
hacks to make things work, such as `hashUncoerced`. I'm not 100% sure
how we were able to eliminate that function and still pass all the
behavior tests, but I'm pretty sure things were still broken without
doing type coercion for every generic function call argument.
After the function prototype is evaluated, it produces a deduplicated
`func_instance` `InternPool.Index` which can then be used for the
generic function call.
Some other nice things made by this simplification are the removal of
`comptime_args_fn_inst` and `preallocated_new_func` from `Sema`, and the
messy logic associated with them.
I have not yet been able to measure the perf of this against master
branch. On one hand, it reduces memory usage and pointer chasing of the
most heavily used `InternPool` Tag - function bodies - but on the other
hand, it does evaluate function prototype expressions more than before.
We will soon find out.
2023-07-03 22:09:30 -07:00
|
|
|
|
|
|
|
|
|
2021-07-13 15:45:08 -07:00
|
|
|
|
2021-08-03 17:29:59 -07:00
|
|
|
|
2021-07-12 15:30:30 -07:00
|
|
|
|
2022-12-30 14:28:04 +02:00
|
|
|
|
rework generic function calls
Abridged summary:
* Move `Module.Fn` into `InternPool`.
* Delete a lot of confusing and problematic `Sema` logic related to
generic function calls.
This commit removes `Module.Fn` and replaces it with two new
`InternPool.Tag` values:
* `func_decl` - corresponding to a function declared in the source
code. This one contains line/column numbers, zir_body_inst, etc.
* `func_instance` - one for each monomorphization of a generic
function. Contains a reference to the `func_decl` from whence the
instantiation came, along with the `comptime` parameter values (or
types in the case of `anytype`)
Since `InternPool` provides deduplication on these values, these fields
are now deleted from `Module`:
* `monomorphed_func_keys`
* `monomorphed_funcs`
* `align_stack_fns`
Instead of these, Sema logic for generic function instantiation now
unconditionally evaluates the function prototype expression for every
generic callsite. This is technically required in order for type
coercions to work. The previous code had some dubious, probably wrong
hacks to make things work, such as `hashUncoerced`. I'm not 100% sure
how we were able to eliminate that function and still pass all the
behavior tests, but I'm pretty sure things were still broken without
doing type coercion for every generic function call argument.
After the function prototype is evaluated, it produces a deduplicated
`func_instance` `InternPool.Index` which can then be used for the
generic function call.
Some other nice things made by this simplification are the removal of
`comptime_args_fn_inst` and `preallocated_new_func` from `Sema`, and the
messy logic associated with them.
I have not yet been able to measure the perf of this against master
branch. On one hand, it reduces memory usage and pointer chasing of the
most heavily used `InternPool` Tag - function bodies - but on the other
hand, it does evaluate function prototype expressions more than before.
We will soon find out.
2023-07-03 22:09:30 -07:00
|
|
|
|
|
|
|
|
|
2022-12-30 14:28:04 +02:00
|
|
|
|
2021-07-12 15:30:30 -07:00
|
|
|
|
|
|
|
|
|
2020-09-13 19:17:58 -07:00
|
|
|
|
rework generic function calls
Abridged summary:
* Move `Module.Fn` into `InternPool`.
* Delete a lot of confusing and problematic `Sema` logic related to
generic function calls.
This commit removes `Module.Fn` and replaces it with two new
`InternPool.Tag` values:
* `func_decl` - corresponding to a function declared in the source
code. This one contains line/column numbers, zir_body_inst, etc.
* `func_instance` - one for each monomorphization of a generic
function. Contains a reference to the `func_decl` from whence the
instantiation came, along with the `comptime` parameter values (or
types in the case of `anytype`)
Since `InternPool` provides deduplication on these values, these fields
are now deleted from `Module`:
* `monomorphed_func_keys`
* `monomorphed_funcs`
* `align_stack_fns`
Instead of these, Sema logic for generic function instantiation now
unconditionally evaluates the function prototype expression for every
generic callsite. This is technically required in order for type
coercions to work. The previous code had some dubious, probably wrong
hacks to make things work, such as `hashUncoerced`. I'm not 100% sure
how we were able to eliminate that function and still pass all the
behavior tests, but I'm pretty sure things were still broken without
doing type coercion for every generic function call argument.
After the function prototype is evaluated, it produces a deduplicated
`func_instance` `InternPool.Index` which can then be used for the
generic function call.
Some other nice things made by this simplification are the removal of
`comptime_args_fn_inst` and `preallocated_new_func` from `Sema`, and the
messy logic associated with them.
I have not yet been able to measure the perf of this against master
branch. On one hand, it reduces memory usage and pointer chasing of the
most heavily used `InternPool` Tag - function bodies - but on the other
hand, it does evaluate function prototype expressions more than before.
We will soon find out.
2023-07-03 22:09:30 -07:00
|
|
|
|
2020-09-13 19:17:58 -07:00
|
|
|
|
2022-04-22 21:30:54 +03:00
|
|
|
|
|
|
|
|
|
2022-09-25 19:51:38 -07:00
|
|
|
|
|
|
|
|
|
2022-10-21 12:42:27 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2022-09-25 19:51:38 -07:00
|
|
|
|
2022-03-09 12:24:18 +02:00
|
|
|
|
2021-09-14 21:58:22 -07:00
|
|
|
|
2021-08-05 23:20:53 -07:00
|
|
|
|
|
|
|
|
|
2021-09-14 21:58:22 -07:00
|
|
|
|
2021-08-03 17:29:59 -07:00
|
|
|
|
|
|
|
|
|
2021-04-27 18:36:12 -07:00
|
|
|
|
2022-10-17 15:19:16 +03:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-05-04 20:30:25 -07:00
|
|
|
|
2022-10-17 15:19:16 +03:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2022-04-22 21:30:54 +03:00
|
|
|
|
rework generic function calls
Abridged summary:
* Move `Module.Fn` into `InternPool`.
* Delete a lot of confusing and problematic `Sema` logic related to
generic function calls.
This commit removes `Module.Fn` and replaces it with two new
`InternPool.Tag` values:
* `func_decl` - corresponding to a function declared in the source
code. This one contains line/column numbers, zir_body_inst, etc.
* `func_instance` - one for each monomorphization of a generic
function. Contains a reference to the `func_decl` from whence the
instantiation came, along with the `comptime` parameter values (or
types in the case of `anytype`)
Since `InternPool` provides deduplication on these values, these fields
are now deleted from `Module`:
* `monomorphed_func_keys`
* `monomorphed_funcs`
* `align_stack_fns`
Instead of these, Sema logic for generic function instantiation now
unconditionally evaluates the function prototype expression for every
generic callsite. This is technically required in order for type
coercions to work. The previous code had some dubious, probably wrong
hacks to make things work, such as `hashUncoerced`. I'm not 100% sure
how we were able to eliminate that function and still pass all the
behavior tests, but I'm pretty sure things were still broken without
doing type coercion for every generic function call argument.
After the function prototype is evaluated, it produces a deduplicated
`func_instance` `InternPool.Index` which can then be used for the
generic function call.
Some other nice things made by this simplification are the removal of
`comptime_args_fn_inst` and `preallocated_new_func` from `Sema`, and the
messy logic associated with them.
I have not yet been able to measure the perf of this against master
branch. On one hand, it reduces memory usage and pointer chasing of the
most heavily used `InternPool` Tag - function bodies - but on the other
hand, it does evaluate function prototype expressions more than before.
We will soon find out.
2023-07-03 22:09:30 -07:00
|
|
|
|
2022-04-22 21:30:54 +03:00
|
|
|
|
2023-05-02 15:01:45 -07:00
|
|
|
|
2022-04-22 21:30:54 +03:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-06-02 01:55:16 +01:00
|
|
|
|
|
|
|
|
|
2023-07-10 10:52:17 -04:00
|
|
|
|
2023-06-02 01:55:16 +01:00
|
|
|
|
2021-09-23 12:17:06 -05:00
|
|
|
|
2021-07-11 16:32:11 -07:00
|
|
|
|
2021-07-15 15:52:06 -07:00
|
|
|
|
|
|
|
|
|
2021-07-12 15:30:30 -07:00
|
|
|
|
2023-07-08 23:39:37 -07:00
|
|
|
|
2021-07-12 15:30:30 -07:00
|
|
|
|
|
|
|
|
|
2023-06-15 13:14:16 +06:00
|
|
|
|
2021-07-11 16:32:11 -07:00
|
|
|
|
2023-07-16 23:17:19 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
rework generic function calls
Abridged summary:
* Move `Module.Fn` into `InternPool`.
* Delete a lot of confusing and problematic `Sema` logic related to
generic function calls.
This commit removes `Module.Fn` and replaces it with two new
`InternPool.Tag` values:
* `func_decl` - corresponding to a function declared in the source
code. This one contains line/column numbers, zir_body_inst, etc.
* `func_instance` - one for each monomorphization of a generic
function. Contains a reference to the `func_decl` from whence the
instantiation came, along with the `comptime` parameter values (or
types in the case of `anytype`)
Since `InternPool` provides deduplication on these values, these fields
are now deleted from `Module`:
* `monomorphed_func_keys`
* `monomorphed_funcs`
* `align_stack_fns`
Instead of these, Sema logic for generic function instantiation now
unconditionally evaluates the function prototype expression for every
generic callsite. This is technically required in order for type
coercions to work. The previous code had some dubious, probably wrong
hacks to make things work, such as `hashUncoerced`. I'm not 100% sure
how we were able to eliminate that function and still pass all the
behavior tests, but I'm pretty sure things were still broken without
doing type coercion for every generic function call argument.
After the function prototype is evaluated, it produces a deduplicated
`func_instance` `InternPool.Index` which can then be used for the
generic function call.
Some other nice things made by this simplification are the removal of
`comptime_args_fn_inst` and `preallocated_new_func` from `Sema`, and the
messy logic associated with them.
I have not yet been able to measure the perf of this against master
branch. On one hand, it reduces memory usage and pointer chasing of the
most heavily used `InternPool` Tag - function bodies - but on the other
hand, it does evaluate function prototype expressions more than before.
We will soon find out.
2023-07-03 22:09:30 -07:00
|
|
|
|
2021-07-11 16:32:11 -07:00
|
|
|
|
2022-03-16 13:31:53 -07:00
|
|
|
|
|
|
|
|
|
2022-03-23 18:45:51 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-07-16 23:17:19 -07:00
|
|
|
|
2022-03-16 13:31:53 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2022-04-02 19:12:08 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2022-03-16 13:31:53 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2022-03-23 18:45:51 -07:00
|
|
|
|
|
|
|
|
|
2023-05-23 19:10:36 -07:00
|
|
|
|
|
|
|
|
|
2022-03-23 18:45:51 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2022-04-02 19:12:08 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2022-03-23 18:45:51 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
rework generic function calls
Abridged summary:
* Move `Module.Fn` into `InternPool`.
* Delete a lot of confusing and problematic `Sema` logic related to
generic function calls.
This commit removes `Module.Fn` and replaces it with two new
`InternPool.Tag` values:
* `func_decl` - corresponding to a function declared in the source
code. This one contains line/column numbers, zir_body_inst, etc.
* `func_instance` - one for each monomorphization of a generic
function. Contains a reference to the `func_decl` from whence the
instantiation came, along with the `comptime` parameter values (or
types in the case of `anytype`)
Since `InternPool` provides deduplication on these values, these fields
are now deleted from `Module`:
* `monomorphed_func_keys`
* `monomorphed_funcs`
* `align_stack_fns`
Instead of these, Sema logic for generic function instantiation now
unconditionally evaluates the function prototype expression for every
generic callsite. This is technically required in order for type
coercions to work. The previous code had some dubious, probably wrong
hacks to make things work, such as `hashUncoerced`. I'm not 100% sure
how we were able to eliminate that function and still pass all the
behavior tests, but I'm pretty sure things were still broken without
doing type coercion for every generic function call argument.
After the function prototype is evaluated, it produces a deduplicated
`func_instance` `InternPool.Index` which can then be used for the
generic function call.
Some other nice things made by this simplification are the removal of
`comptime_args_fn_inst` and `preallocated_new_func` from `Sema`, and the
messy logic associated with them.
I have not yet been able to measure the perf of this against master
branch. On one hand, it reduces memory usage and pointer chasing of the
most heavily used `InternPool` Tag - function bodies - but on the other
hand, it does evaluate function prototype expressions more than before.
We will soon find out.
2023-07-03 22:09:30 -07:00
|
|
|
|
2021-07-11 16:32:11 -07:00
|
|
|
|
2022-11-27 01:07:35 -07:00
|
|
|
|
2021-07-11 16:32:11 -07:00
|
|
|
|
2020-09-13 19:17:58 -07:00
|
|
|
|
|
|
|
|
|
2022-04-19 21:51:08 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-06-03 15:39:26 -05:00
|
|
|
|
2020-09-13 19:17:58 -07:00
|
|
|
|
2022-12-12 13:16:09 +02:00
|
|
|
|
2023-09-25 04:54:07 +02:00
|
|
|
|
|
|
|
|
|
2022-12-12 13:16:09 +02:00
|
|
|
|
2021-04-26 20:41:07 -07:00
|
|
|
|
2022-04-19 21:51:08 -07:00
|
|
|
|
2021-06-03 15:39:26 -05:00
|
|
|
|
2021-04-26 20:41:07 -07:00
|
|
|
|
2021-01-05 17:33:31 -07:00
|
|
|
|
2022-04-19 21:51:08 -07:00
|
|
|
|
2020-09-13 19:17:58 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-05-10 17:21:22 -07:00
|
|
|
|
2023-07-06 15:27:47 -07:00
|
|
|
|
2023-05-08 16:52:59 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-07-06 15:27:47 -07:00
|
|
|
|
2023-05-08 16:52:59 -07:00
|
|
|
|
|
|
|
|
|
2021-10-22 15:53:59 -07:00
|
|
|
|
|
|
|
|
|
2023-05-08 16:52:59 -07:00
|
|
|
|
2021-10-22 15:53:59 -07:00
|
|
|
|
2023-09-14 19:59:58 -07:00
|
|
|
|
2022-04-19 21:51:08 -07:00
|
|
|
|
2023-07-06 15:27:47 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2022-04-19 21:51:08 -07:00
|
|
|
|
2021-10-01 12:05:12 -05:00
|
|
|
|
2021-04-08 20:37:19 -07:00
|
|
|
|
2021-05-01 21:57:52 -07:00
|
|
|
|
2021-04-27 18:36:12 -07:00
|
|
|
|
2021-05-11 14:17:52 -07:00
|
|
|
|
2021-04-27 18:36:12 -07:00
|
|
|
|
|
|
|
|
|
2023-06-20 14:02:04 -04:00
|
|
|
|
compiler: eliminate Decl.value_arena and Sema.perm_arena
The main motivation for this commit is eliminating Decl.value_arena.
Everything else is dominoes.
Decl.name used to be stored in the GPA, now it is stored in InternPool.
It ended up being simpler to migrate other strings to be interned as
well, such as struct field names, union field names, and a few others.
This ended up requiring a big diff, sorry about that. But the changes
are pretty nice, we finally start to take advantage of InternPool's
existence.
global_error_set and error_name_list are simplified. Now it is a single
ArrayHashMap(NullTerminatedString, void) and the index is the error tag
value.
Module.tmp_hack_arena is re-introduced (it was removed in
eeff407941560ce8eb5b737b2436dfa93cfd3a0c) in order to deal with
comptime_args, optimized_order, and struct and union fields. After
structs and unions get moved into InternPool properly, tmp_hack_arena
can be deleted again.
2023-06-01 23:46:04 -07:00
|
|
|
|
2021-10-22 15:53:59 -07:00
|
|
|
|
2020-09-13 19:17:58 -07:00
|
|
|
|
|
|
|
|
|
2021-05-05 13:16:14 -07:00
|
|
|
|
2021-09-23 12:17:06 -05:00
|
|
|
|
2020-09-13 19:17:58 -07:00
|
|
|
|
|
|
|
|
|
2021-04-27 18:36:12 -07:00
|
|
|
|
2021-09-02 14:46:31 +02:00
|
|
|
|
2021-04-28 16:55:22 -07:00
|
|
|
|
stage2: garbage collect unused anon decls
After this change, the frontend and backend cooperate to keep track of
which Decls are actually emitted into the machine code. When any backend
sees a `decl_ref` Value, it must mark the corresponding Decl `alive`
field to true.
This prevents unused comptime data from spilling into the output object
files. For example, if you do an `inline for` loop, previously, any
intermediate value calculations would have gone into the object file.
Now they are garbage collected immediately after the owner Decl has its
machine code generated.
In the frontend, when it is time to send a Decl to the linker, if it has
not been marked "alive" then it is deleted instead.
Additional improvements:
* Resolve type ABI layouts after successful semantic analysis of a
Decl. This is needed so that the backend has access to struct fields.
* Sema: fix incorrect logic in resolveMaybeUndefVal. It should return
"not comptime known" instead of a compile error for global variables.
* `Value.pointerDeref` now returns `null` in the case that the pointer
deref cannot happen at compile-time. This is true for global
variables, for example. Another example is if a comptime known
pointer has a hard coded address value.
* Binary arithmetic sets the requireRuntimeBlock source location to the
lhs_src or rhs_src as appropriate instead of on the operator node.
* Fix LLVM codegen for slice_elem_val which had the wrong logic for
when the operand was not a pointer.
As noted in the comment in the implementation of deleteUnusedDecl, a
future improvement will be to rework the frontend/linker interface to
remove the frontend's responsibility of calling allocateDeclIndexes.
I discovered some issues with the plan9 linker backend that are related
to this, and worked around them for now.
2021-07-29 19:30:37 -07:00
|
|
|
|
2023-01-10 21:54:58 +02:00
|
|
|
|
2023-07-06 15:27:47 -07:00
|
|
|
|
2021-09-23 12:17:06 -05:00
|
|
|
|
2023-07-06 15:27:47 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2020-09-13 19:17:58 -07:00
|
|
|
|
|
|
|
|
|
compiler: eliminate Decl.value_arena and Sema.perm_arena
The main motivation for this commit is eliminating Decl.value_arena.
Everything else is dominoes.
Decl.name used to be stored in the GPA, now it is stored in InternPool.
It ended up being simpler to migrate other strings to be interned as
well, such as struct field names, union field names, and a few others.
This ended up requiring a big diff, sorry about that. But the changes
are pretty nice, we finally start to take advantage of InternPool's
existence.
global_error_set and error_name_list are simplified. Now it is a single
ArrayHashMap(NullTerminatedString, void) and the index is the error tag
value.
Module.tmp_hack_arena is re-introduced (it was removed in
eeff407941560ce8eb5b737b2436dfa93cfd3a0c) in order to deal with
comptime_args, optimized_order, and struct and union fields. After
structs and unions get moved into InternPool properly, tmp_hack_arena
can be deleted again.
2023-06-01 23:46:04 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
stage2: *WIP*: rework ZIR memory layout; overhaul source locations
The memory layout for ZIR instructions is completely reworked. See
zir.zig for those changes. Some new types:
* `zir.Code`: a "finished" set of ZIR instructions. Instead of allocating
each instruction independently, there is now a Tag and 8 bytes of
data available for all ZIR instructions. Small instructions fit
within these 8 bytes; larger ones use 4 bytes for an index into
`extra`. There is also `string_bytes` so that we can have 4 byte
references to strings. `zir.Inst.Tag` describes how to interpret
those 8 bytes of data.
- This is shared by all `Block` scopes.
* `Module.WipZirCode`: represents an in-progress `zir.Code`. In this
structure, the arrays are mutable, and get resized as we add/delete
things. There is extra state to keep track of things. This struct is
stored on the stack. Once it is finished, it produces an immutable
`zir.Code`, which will remain on the heap for the duration of a
function's existence.
- This is shared by all `GenZir` scopes.
* `Sema`: represents in-progress semantic analysis of a `zir.Code`.
This data is stored on the stack and is shared among all `Block`
scopes. It is now the main "self" argument to everything in the file
that was previously named `zir_sema.zig`.
Additionally, I moved some logic that was in `Module` into here.
`Module.Fn` now stores its parameter names inside the `zir.Code`,
instead of inside ZIR instructions. When the TZIR memory layout
reworking time comes, codegen will be able to reference this data
directly instead of duplicating it.
astgen.zig is (so far) almost entirely untouched, but nearly all of it
will need to be reworked to adhere to this new memory layout structure.
I have no benchmarks to report yet, as I am still working through
compile errors and fixing various things that I broke in this branch.
Overhaul of Source Locations:
Previously we used `usize` everywhere to mean byte offset, but sometimes
also mean other stuff. This was error prone and also made us do
unnecessary work, and store unnecessary bytes in memory.
Now there are more types involved into source locations, and more ways
to describe a source location.
* AllErrors.Message: embrace the assumption that files always have less
than 2 << 32 bytes.
* SrcLoc gets more complicated, to model more complicated source
locations.
* Introduce LazySrcLoc, which can model interesting source locations
with very little stored state. Useful for avoiding doing unnecessary
work when no compile errors occur.
Also, previously, we had `src: usize` on every ZIR instruction. This is
no longer the case. Each instruction now determines whether it even cares
about source location, and if so, how that source location is stored.
This requires more careful work inside `Sema`, but it results in fewer
bytes stored on the heap, without compromising accuracy and power of
compile error messages.
Miscellaneous:
* std.zig: string literals have more helpful result values for
reporting errors. There is now a lower level API and a higher level
API.
- side note: I noticed that the string literal logic needs some love.
There is some unnecessarily hacky code there.
* cut & pasted some TZIR logic that was in zir.zig to ir.zig. This
probably broke stuff and needs to get fixed.
* Removed type/Enum.zig, type/Union.zig, and type/Struct.zig. I don't
think this quite how this code will be organized. Need some more
careful planning about how to implement structs, unions, enums. They
need to be independent Decls, just like a top level function.
2021-03-15 23:38:38 -07:00
|
|
|
|
2023-06-22 18:46:56 +01:00
|
|
|
|
compiler: eliminate Decl.value_arena and Sema.perm_arena
The main motivation for this commit is eliminating Decl.value_arena.
Everything else is dominoes.
Decl.name used to be stored in the GPA, now it is stored in InternPool.
It ended up being simpler to migrate other strings to be interned as
well, such as struct field names, union field names, and a few others.
This ended up requiring a big diff, sorry about that. But the changes
are pretty nice, we finally start to take advantage of InternPool's
existence.
global_error_set and error_name_list are simplified. Now it is a single
ArrayHashMap(NullTerminatedString, void) and the index is the error tag
value.
Module.tmp_hack_arena is re-introduced (it was removed in
eeff407941560ce8eb5b737b2436dfa93cfd3a0c) in order to deal with
comptime_args, optimized_order, and struct and union fields. After
structs and unions get moved into InternPool properly, tmp_hack_arena
can be deleted again.
2023-06-01 23:46:04 -07:00
|
|
|
|
2020-09-13 19:17:58 -07:00
|
|
|
|
compiler: eliminate Decl.value_arena and Sema.perm_arena
The main motivation for this commit is eliminating Decl.value_arena.
Everything else is dominoes.
Decl.name used to be stored in the GPA, now it is stored in InternPool.
It ended up being simpler to migrate other strings to be interned as
well, such as struct field names, union field names, and a few others.
This ended up requiring a big diff, sorry about that. But the changes
are pretty nice, we finally start to take advantage of InternPool's
existence.
global_error_set and error_name_list are simplified. Now it is a single
ArrayHashMap(NullTerminatedString, void) and the index is the error tag
value.
Module.tmp_hack_arena is re-introduced (it was removed in
eeff407941560ce8eb5b737b2436dfa93cfd3a0c) in order to deal with
comptime_args, optimized_order, and struct and union fields. After
structs and unions get moved into InternPool properly, tmp_hack_arena
can be deleted again.
2023-06-01 23:46:04 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2020-09-13 19:17:58 -07:00
|
|
|
|
|
|
|
|
|
2023-01-24 13:35:10 +02:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2022-04-19 21:51:08 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
stage2: `zig test` now works with the LLVM backend
Frontend improvements:
* When compiling in `zig test` mode, put a task on the work queue to
analyze the main package root file. Normally, start code does
`_ = import("root");` to make Zig analyze the user's code, however in
the case of `zig test`, the root source file is the test runner.
Without this change, no tests are picked up.
* In the main pipeline, once semantic analysis is finished, if there
are no compile errors, populate the `test_functions` Decl with the
set of test functions picked up from semantic analysis.
* Value: add `array` and `slice` Tags.
LLVM backend improvements:
* Fix incremental updates of globals. Previously the
value of a global would not get replaced with a new value.
* Fix LLVM type of arrays. They were incorrectly sending
the ABI size as the element count.
* Remove the FuncGen parameter from genTypedValue. This function is for
generating global constants and there is no function available when
it is being called.
- The `ref_val` case is now commented out. I'd like to eliminate
`ref_val` as one of the possible Value Tags. Instead it should
always be done via `decl_ref`.
* Implement constant value generation for slices, arrays, and structs.
* Constant value generation for functions supports the `decl_ref` tag.
2021-07-27 14:06:42 -07:00
|
|
|
|
|
|
|
|
|
2021-10-01 12:05:12 -05:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-05-08 16:52:59 -07:00
|
|
|
|
2023-09-14 19:59:58 -07:00
|
|
|
|
2021-10-01 12:05:12 -05:00
|
|
|
|
2022-04-19 21:51:08 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-06-12 01:44:12 -04:00
|
|
|
|
2023-06-15 13:14:16 +06:00
|
|
|
|
2023-06-12 01:44:12 -04:00
|
|
|
|
2023-07-06 15:27:47 -07:00
|
|
|
|
2022-04-19 21:51:08 -07:00
|
|
|
|
stage2: `zig test` now works with the LLVM backend
Frontend improvements:
* When compiling in `zig test` mode, put a task on the work queue to
analyze the main package root file. Normally, start code does
`_ = import("root");` to make Zig analyze the user's code, however in
the case of `zig test`, the root source file is the test runner.
Without this change, no tests are picked up.
* In the main pipeline, once semantic analysis is finished, if there
are no compile errors, populate the `test_functions` Decl with the
set of test functions picked up from semantic analysis.
* Value: add `array` and `slice` Tags.
LLVM backend improvements:
* Fix incremental updates of globals. Previously the
value of a global would not get replaced with a new value.
* Fix LLVM type of arrays. They were incorrectly sending
the ABI size as the element count.
* Remove the FuncGen parameter from genTypedValue. This function is for
generating global constants and there is no function available when
it is being called.
- The `ref_val` case is now commented out. I'd like to eliminate
`ref_val` as one of the possible Value Tags. Instead it should
always be done via `decl_ref`.
* Implement constant value generation for slices, arrays, and structs.
* Constant value generation for functions supports the `decl_ref` tag.
2021-07-27 14:06:42 -07:00
|
|
|
|
|
|
|
|
|
2022-04-19 21:51:08 -07:00
|
|
|
|
stage2: `zig test` now works with the LLVM backend
Frontend improvements:
* When compiling in `zig test` mode, put a task on the work queue to
analyze the main package root file. Normally, start code does
`_ = import("root");` to make Zig analyze the user's code, however in
the case of `zig test`, the root source file is the test runner.
Without this change, no tests are picked up.
* In the main pipeline, once semantic analysis is finished, if there
are no compile errors, populate the `test_functions` Decl with the
set of test functions picked up from semantic analysis.
* Value: add `array` and `slice` Tags.
LLVM backend improvements:
* Fix incremental updates of globals. Previously the
value of a global would not get replaced with a new value.
* Fix LLVM type of arrays. They were incorrectly sending
the ABI size as the element count.
* Remove the FuncGen parameter from genTypedValue. This function is for
generating global constants and there is no function available when
it is being called.
- The `ref_val` case is now commented out. I'd like to eliminate
`ref_val` as one of the possible Value Tags. Instead it should
always be done via `decl_ref`.
* Implement constant value generation for slices, arrays, and structs.
* Constant value generation for functions supports the `decl_ref` tag.
2021-07-27 14:06:42 -07:00
|
|
|
|
2022-04-19 21:51:08 -07:00
|
|
|
|
|
|
|
|
|
stage2: `zig test` now works with the LLVM backend
Frontend improvements:
* When compiling in `zig test` mode, put a task on the work queue to
analyze the main package root file. Normally, start code does
`_ = import("root");` to make Zig analyze the user's code, however in
the case of `zig test`, the root source file is the test runner.
Without this change, no tests are picked up.
* In the main pipeline, once semantic analysis is finished, if there
are no compile errors, populate the `test_functions` Decl with the
set of test functions picked up from semantic analysis.
* Value: add `array` and `slice` Tags.
LLVM backend improvements:
* Fix incremental updates of globals. Previously the
value of a global would not get replaced with a new value.
* Fix LLVM type of arrays. They were incorrectly sending
the ABI size as the element count.
* Remove the FuncGen parameter from genTypedValue. This function is for
generating global constants and there is no function available when
it is being called.
- The `ref_val` case is now commented out. I'd like to eliminate
`ref_val` as one of the possible Value Tags. Instead it should
always be done via `decl_ref`.
* Implement constant value generation for slices, arrays, and structs.
* Constant value generation for functions supports the `decl_ref` tag.
2021-07-27 14:06:42 -07:00
|
|
|
|
compiler: eliminate Decl.value_arena and Sema.perm_arena
The main motivation for this commit is eliminating Decl.value_arena.
Everything else is dominoes.
Decl.name used to be stored in the GPA, now it is stored in InternPool.
It ended up being simpler to migrate other strings to be interned as
well, such as struct field names, union field names, and a few others.
This ended up requiring a big diff, sorry about that. But the changes
are pretty nice, we finally start to take advantage of InternPool's
existence.
global_error_set and error_name_list are simplified. Now it is a single
ArrayHashMap(NullTerminatedString, void) and the index is the error tag
value.
Module.tmp_hack_arena is re-introduced (it was removed in
eeff407941560ce8eb5b737b2436dfa93cfd3a0c) in order to deal with
comptime_args, optimized_order, and struct and union fields. After
structs and unions get moved into InternPool properly, tmp_hack_arena
can be deleted again.
2023-06-01 23:46:04 -07:00
|
|
|
|
2023-05-28 02:41:22 -04:00
|
|
|
|
2023-05-26 21:22:34 -04:00
|
|
|
|
2021-05-10 22:50:00 -07:00
|
|
|
|
2022-04-19 21:51:08 -07:00
|
|
|
|
2021-05-08 14:34:30 -07:00
|
|
|
|
2022-04-19 21:51:08 -07:00
|
|
|
|
|
|
|
|
|
2021-04-27 18:36:12 -07:00
|
|
|
|
|
|
|
|
|
2023-06-20 14:02:04 -04:00
|
|
|
|
compiler: eliminate Decl.value_arena and Sema.perm_arena
The main motivation for this commit is eliminating Decl.value_arena.
Everything else is dominoes.
Decl.name used to be stored in the GPA, now it is stored in InternPool.
It ended up being simpler to migrate other strings to be interned as
well, such as struct field names, union field names, and a few others.
This ended up requiring a big diff, sorry about that. But the changes
are pretty nice, we finally start to take advantage of InternPool's
existence.
global_error_set and error_name_list are simplified. Now it is a single
ArrayHashMap(NullTerminatedString, void) and the index is the error tag
value.
Module.tmp_hack_arena is re-introduced (it was removed in
eeff407941560ce8eb5b737b2436dfa93cfd3a0c) in order to deal with
comptime_args, optimized_order, and struct and union fields. After
structs and unions get moved into InternPool properly, tmp_hack_arena
can be deleted again.
2023-06-01 23:46:04 -07:00
|
|
|
|
2021-04-27 18:36:12 -07:00
|
|
|
|
2020-09-13 19:17:58 -07:00
|
|
|
|
stage2: *WIP*: rework ZIR memory layout; overhaul source locations
The memory layout for ZIR instructions is completely reworked. See
zir.zig for those changes. Some new types:
* `zir.Code`: a "finished" set of ZIR instructions. Instead of allocating
each instruction independently, there is now a Tag and 8 bytes of
data available for all ZIR instructions. Small instructions fit
within these 8 bytes; larger ones use 4 bytes for an index into
`extra`. There is also `string_bytes` so that we can have 4 byte
references to strings. `zir.Inst.Tag` describes how to interpret
those 8 bytes of data.
- This is shared by all `Block` scopes.
* `Module.WipZirCode`: represents an in-progress `zir.Code`. In this
structure, the arrays are mutable, and get resized as we add/delete
things. There is extra state to keep track of things. This struct is
stored on the stack. Once it is finished, it produces an immutable
`zir.Code`, which will remain on the heap for the duration of a
function's existence.
- This is shared by all `GenZir` scopes.
* `Sema`: represents in-progress semantic analysis of a `zir.Code`.
This data is stored on the stack and is shared among all `Block`
scopes. It is now the main "self" argument to everything in the file
that was previously named `zir_sema.zig`.
Additionally, I moved some logic that was in `Module` into here.
`Module.Fn` now stores its parameter names inside the `zir.Code`,
instead of inside ZIR instructions. When the TZIR memory layout
reworking time comes, codegen will be able to reference this data
directly instead of duplicating it.
astgen.zig is (so far) almost entirely untouched, but nearly all of it
will need to be reworked to adhere to this new memory layout structure.
I have no benchmarks to report yet, as I am still working through
compile errors and fixing various things that I broke in this branch.
Overhaul of Source Locations:
Previously we used `usize` everywhere to mean byte offset, but sometimes
also mean other stuff. This was error prone and also made us do
unnecessary work, and store unnecessary bytes in memory.
Now there are more types involved into source locations, and more ways
to describe a source location.
* AllErrors.Message: embrace the assumption that files always have less
than 2 << 32 bytes.
* SrcLoc gets more complicated, to model more complicated source
locations.
* Introduce LazySrcLoc, which can model interesting source locations
with very little stored state. Useful for avoiding doing unnecessary
work when no compile errors occur.
Also, previously, we had `src: usize` on every ZIR instruction. This is
no longer the case. Each instruction now determines whether it even cares
about source location, and if so, how that source location is stored.
This requires more careful work inside `Sema`, but it results in fewer
bytes stored on the heap, without compromising accuracy and power of
compile error messages.
Miscellaneous:
* std.zig: string literals have more helpful result values for
reporting errors. There is now a lower level API and a higher level
API.
- side note: I noticed that the string literal logic needs some love.
There is some unnecessarily hacky code there.
* cut & pasted some TZIR logic that was in zir.zig to ir.zig. This
probably broke stuff and needs to get fixed.
* Removed type/Enum.zig, type/Union.zig, and type/Struct.zig. I don't
think this quite how this code will be organized. Need some more
careful planning about how to implement structs, unions, enums. They
need to be independent Decls, just like a top level function.
2021-03-15 23:38:38 -07:00
|
|
|
|
stage2: type declarations ZIR encode AnonNameStrategy
which can be either parent, func, or anon. Here's the enum reproduced in
the commit message for convenience:
```zig
pub const NameStrategy = enum(u2) {
/// Use the same name as the parent declaration name.
/// e.g. `const Foo = struct {...};`.
parent,
/// Use the name of the currently executing comptime function call,
/// with the current parameters. e.g. `ArrayList(i32)`.
func,
/// Create an anonymous name for this declaration.
/// Like this: "ParentDeclName_struct_69"
anon,
};
```
With this information in the ZIR, a future commit can improve the
names of structs, unions, enums, and opaques.
In order to accomplish this, the following ZIR instruction forms were
removed and replaced with Extended op codes:
* struct_decl
* struct_decl_packed
* struct_decl_extern
* union_decl
* union_decl_packed
* union_decl_extern
* enum_decl
* enum_decl_nonexhaustive
By being extended opcodes, one more u32 is needed, however we more than
make up for it by repurposing the 16 "small" bits to provide shorter
encodings for when decls_len == 0, fields_len == 0, a source node is not
provided, etc. There tends to be no downside, and in fact sometimes
upsides, to using an extended op code when there is a need for flag
bits, which is the case for all three of these. Likewise, the container
layout can be encoded in these bits rather than into the opcode.
The following 4 ZIR instructions were added, netting a total of 4 freed
up ZIR enum tags for future use:
* opaque_decl_anon
* opaque_decl_func
* error_set_decl_anon
* error_set_decl_func
This is so that opaques and error sets can have the same name hint as
structs, enums, and unions.
`std.builtin.ContainerLayout` gets an explicit integer tag type so that
it can be used inside packed structs.
This commit also makes `Module.Namespace` use a separate set for
anonymous decls, thus allowing anonymous decls to share the same
`Decl.name` as their owner `Decl` objects.
2021-05-10 21:34:43 -07:00
|
|
|
|
|
|
|
|
|
2021-04-02 21:06:09 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-01-16 22:51:01 -07:00
|
|
|
|
2023-05-31 07:15:52 -04:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-01-16 22:51:01 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-04-02 21:06:09 -07:00
|
|
|
|
2021-01-16 22:51:01 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
stage2: *WIP*: rework ZIR memory layout; overhaul source locations
The memory layout for ZIR instructions is completely reworked. See
zir.zig for those changes. Some new types:
* `zir.Code`: a "finished" set of ZIR instructions. Instead of allocating
each instruction independently, there is now a Tag and 8 bytes of
data available for all ZIR instructions. Small instructions fit
within these 8 bytes; larger ones use 4 bytes for an index into
`extra`. There is also `string_bytes` so that we can have 4 byte
references to strings. `zir.Inst.Tag` describes how to interpret
those 8 bytes of data.
- This is shared by all `Block` scopes.
* `Module.WipZirCode`: represents an in-progress `zir.Code`. In this
structure, the arrays are mutable, and get resized as we add/delete
things. There is extra state to keep track of things. This struct is
stored on the stack. Once it is finished, it produces an immutable
`zir.Code`, which will remain on the heap for the duration of a
function's existence.
- This is shared by all `GenZir` scopes.
* `Sema`: represents in-progress semantic analysis of a `zir.Code`.
This data is stored on the stack and is shared among all `Block`
scopes. It is now the main "self" argument to everything in the file
that was previously named `zir_sema.zig`.
Additionally, I moved some logic that was in `Module` into here.
`Module.Fn` now stores its parameter names inside the `zir.Code`,
instead of inside ZIR instructions. When the TZIR memory layout
reworking time comes, codegen will be able to reference this data
directly instead of duplicating it.
astgen.zig is (so far) almost entirely untouched, but nearly all of it
will need to be reworked to adhere to this new memory layout structure.
I have no benchmarks to report yet, as I am still working through
compile errors and fixing various things that I broke in this branch.
Overhaul of Source Locations:
Previously we used `usize` everywhere to mean byte offset, but sometimes
also mean other stuff. This was error prone and also made us do
unnecessary work, and store unnecessary bytes in memory.
Now there are more types involved into source locations, and more ways
to describe a source location.
* AllErrors.Message: embrace the assumption that files always have less
than 2 << 32 bytes.
* SrcLoc gets more complicated, to model more complicated source
locations.
* Introduce LazySrcLoc, which can model interesting source locations
with very little stored state. Useful for avoiding doing unnecessary
work when no compile errors occur.
Also, previously, we had `src: usize` on every ZIR instruction. This is
no longer the case. Each instruction now determines whether it even cares
about source location, and if so, how that source location is stored.
This requires more careful work inside `Sema`, but it results in fewer
bytes stored on the heap, without compromising accuracy and power of
compile error messages.
Miscellaneous:
* std.zig: string literals have more helpful result values for
reporting errors. There is now a lower level API and a higher level
API.
- side note: I noticed that the string literal logic needs some love.
There is some unnecessarily hacky code there.
* cut & pasted some TZIR logic that was in zir.zig to ir.zig. This
probably broke stuff and needs to get fixed.
* Removed type/Enum.zig, type/Union.zig, and type/Struct.zig. I don't
think this quite how this code will be organized. Need some more
careful planning about how to implement structs, unions, enums. They
need to be independent Decls, just like a top level function.
2021-03-15 23:38:38 -07:00
|
|
|
|
|
|
|
|
|
2020-09-13 19:17:58 -07:00
|
|
|
|
|
|
|
|
|
2023-08-09 13:39:34 -05:00
|
|
|
|
stage2: *WIP*: rework ZIR memory layout; overhaul source locations
The memory layout for ZIR instructions is completely reworked. See
zir.zig for those changes. Some new types:
* `zir.Code`: a "finished" set of ZIR instructions. Instead of allocating
each instruction independently, there is now a Tag and 8 bytes of
data available for all ZIR instructions. Small instructions fit
within these 8 bytes; larger ones use 4 bytes for an index into
`extra`. There is also `string_bytes` so that we can have 4 byte
references to strings. `zir.Inst.Tag` describes how to interpret
those 8 bytes of data.
- This is shared by all `Block` scopes.
* `Module.WipZirCode`: represents an in-progress `zir.Code`. In this
structure, the arrays are mutable, and get resized as we add/delete
things. There is extra state to keep track of things. This struct is
stored on the stack. Once it is finished, it produces an immutable
`zir.Code`, which will remain on the heap for the duration of a
function's existence.
- This is shared by all `GenZir` scopes.
* `Sema`: represents in-progress semantic analysis of a `zir.Code`.
This data is stored on the stack and is shared among all `Block`
scopes. It is now the main "self" argument to everything in the file
that was previously named `zir_sema.zig`.
Additionally, I moved some logic that was in `Module` into here.
`Module.Fn` now stores its parameter names inside the `zir.Code`,
instead of inside ZIR instructions. When the TZIR memory layout
reworking time comes, codegen will be able to reference this data
directly instead of duplicating it.
astgen.zig is (so far) almost entirely untouched, but nearly all of it
will need to be reworked to adhere to this new memory layout structure.
I have no benchmarks to report yet, as I am still working through
compile errors and fixing various things that I broke in this branch.
Overhaul of Source Locations:
Previously we used `usize` everywhere to mean byte offset, but sometimes
also mean other stuff. This was error prone and also made us do
unnecessary work, and store unnecessary bytes in memory.
Now there are more types involved into source locations, and more ways
to describe a source location.
* AllErrors.Message: embrace the assumption that files always have less
than 2 << 32 bytes.
* SrcLoc gets more complicated, to model more complicated source
locations.
* Introduce LazySrcLoc, which can model interesting source locations
with very little stored state. Useful for avoiding doing unnecessary
work when no compile errors occur.
Also, previously, we had `src: usize` on every ZIR instruction. This is
no longer the case. Each instruction now determines whether it even cares
about source location, and if so, how that source location is stored.
This requires more careful work inside `Sema`, but it results in fewer
bytes stored on the heap, without compromising accuracy and power of
compile error messages.
Miscellaneous:
* std.zig: string literals have more helpful result values for
reporting errors. There is now a lower level API and a higher level
API.
- side note: I noticed that the string literal logic needs some love.
There is some unnecessarily hacky code there.
* cut & pasted some TZIR logic that was in zir.zig to ir.zig. This
probably broke stuff and needs to get fixed.
* Removed type/Enum.zig, type/Union.zig, and type/Struct.zig. I don't
think this quite how this code will be organized. Need some more
careful planning about how to implement structs, unions, enums. They
need to be independent Decls, just like a top level function.
2021-03-15 23:38:38 -07:00
|
|
|
|
2020-09-13 19:17:58 -07:00
|
|
|
|
2021-04-28 16:55:22 -07:00
|
|
|
|
2021-10-01 17:45:13 -05:00
|
|
|
|
2021-04-28 16:55:22 -07:00
|
|
|
|
2021-05-11 17:34:13 -07:00
|
|
|
|
2021-04-28 16:55:22 -07:00
|
|
|
|
2021-11-09 18:27:12 -07:00
|
|
|
|
|
|
|
|
|
2021-06-03 15:39:26 -05:00
|
|
|
|
|
|
|
|
|
2021-04-28 16:55:22 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-05-02 17:08:19 -07:00
|
|
|
|
|
|
|
|
|
2023-05-28 04:31:56 +01:00
|
|
|
|
2021-05-02 17:08:19 -07:00
|
|
|
|
2023-05-28 04:31:56 +01:00
|
|
|
|
2021-05-02 17:08:19 -07:00
|
|
|
|
2023-05-28 04:31:56 +01:00
|
|
|
|
2021-05-02 17:08:19 -07:00
|
|
|
|
2023-05-28 04:31:56 +01:00
|
|
|
|
2023-05-26 04:10:51 +01:00
|
|
|
|
2023-05-28 04:31:56 +01:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-05-02 17:08:19 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-05-08 16:52:59 -07:00
|
|
|
|
2021-05-02 17:08:19 -07:00
|
|
|
|
|
|
|
|
|
2023-05-28 04:31:56 +01:00
|
|
|
|
2021-05-02 17:08:19 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-05-08 16:52:59 -07:00
|
|
|
|
|
|
|
|
|
2021-05-17 19:11:11 -07:00
|
|
|
|
|
|
|
|
|
2023-05-08 16:52:59 -07:00
|
|
|
|
2021-05-17 19:11:11 -07:00
|
|
|
|
2022-06-07 17:48:53 -07:00
|
|
|
|
2021-05-17 19:11:11 -07:00
|
|
|
|
2021-05-02 17:08:19 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-08-30 19:22:04 -07:00
|
|
|
|
2021-05-02 17:08:19 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-05-28 04:31:56 +01:00
|
|
|
|
2023-01-09 14:59:19 +00:00
|
|
|
|
2023-05-26 04:10:51 +01:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-05-28 04:31:56 +01:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-05-02 17:08:19 -07:00
|
|
|
|
2023-05-26 04:10:51 +01:00
|
|
|
|
2021-05-02 17:08:19 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-05-28 04:31:56 +01:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-05-02 17:08:19 -07:00
|
|
|
|
2023-05-28 04:31:56 +01:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-05-02 17:08:19 -07:00
|
|
|
|
2023-05-28 04:31:56 +01:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-05-02 17:08:19 -07:00
|
|
|
|
2023-05-28 04:31:56 +01:00
|
|
|
|
2021-05-02 17:08:19 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-05-28 04:31:56 +01:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-05-02 17:08:19 -07:00
|
|
|
|
|
|
|
|
|
2021-05-03 11:46:02 -07:00
|
|
|
|
2021-07-09 20:43:19 +08:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-05-10 16:26:54 +03:00
|
|
|
|
2021-07-09 20:43:19 +08:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-05-08 16:52:59 -07:00
|
|
|
|
2021-07-09 20:43:19 +08:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-05-08 16:52:59 -07:00
|
|
|
|
2021-07-09 20:43:19 +08:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-05-07 22:58:15 +02:00
|
|
|
|
|
|
|
|
|
2021-07-09 20:43:19 +08:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-09-14 21:58:22 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-07-09 20:43:19 +08:00
|
|
|
|
|
|
|
|
|
2023-05-08 16:52:59 -07:00
|
|
|
|
2021-07-09 20:43:19 +08:00
|
|
|
|
|
|
|
|
|
2023-05-08 16:52:59 -07:00
|
|
|
|
2021-07-09 20:43:19 +08:00
|
|
|
|
2022-06-07 17:48:53 -07:00
|
|
|
|
2021-07-09 20:43:19 +08:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
stage2: type system treats fn ptr and body separately
This commit updates stage2 to enforce the property that the syntax
`fn()void` is a function *body* not a *pointer*. To get a pointer, the
syntax `*const fn()void` is required.
ZIR puts function alignment into the func instruction rather than the
decl because this way it makes it into function types. LLVM backend
respects function alignments.
Struct and Union have methods `fieldSrcLoc` to help look up source
locations of their fields. These trigger full loading, tokenization, and
parsing of source files, so should only be called once it is confirmed
that an error message needs to be printed.
There are some nice new error hints for explaining why a type is
required to be comptime, particularly for structs that contain function
body types.
`Type.requiresComptime` is now moved into Sema because it can fail and
might need to trigger field type resolution. Comptime pointer loading
takes into account types that do not have a well-defined memory layout
and does not try to compute a byte offset for them.
`fn()void` syntax no longer secretly makes a pointer. You get a function
body type, which requires comptime. However a pointer to a function body
can be runtime known (obviously).
Compile errors that report "expected pointer, found ..." are factored
out into convenience functions `checkPtrOperand` and `checkPtrType` and
have a note about function pointers.
Implemented `Value.hash` for functions, enum literals, and undefined values.
stage1 is not updated to this (yet?), so some workarounds and disabled
tests are needed to keep everything working. Should we update stage1 to
these new type semantics? Yes probably because I don't want to add too
much conditional compilation logic in the std lib for the different
backends.
2022-01-21 00:49:58 -07:00
|
|
|
|
|
|
|
|
|
2023-01-05 21:37:30 +02:00
|
|
|
|
stage2: type system treats fn ptr and body separately
This commit updates stage2 to enforce the property that the syntax
`fn()void` is a function *body* not a *pointer*. To get a pointer, the
syntax `*const fn()void` is required.
ZIR puts function alignment into the func instruction rather than the
decl because this way it makes it into function types. LLVM backend
respects function alignments.
Struct and Union have methods `fieldSrcLoc` to help look up source
locations of their fields. These trigger full loading, tokenization, and
parsing of source files, so should only be called once it is confirmed
that an error message needs to be printed.
There are some nice new error hints for explaining why a type is
required to be comptime, particularly for structs that contain function
body types.
`Type.requiresComptime` is now moved into Sema because it can fail and
might need to trigger field type resolution. Comptime pointer loading
takes into account types that do not have a well-defined memory layout
and does not try to compute a byte offset for them.
`fn()void` syntax no longer secretly makes a pointer. You get a function
body type, which requires comptime. However a pointer to a function body
can be runtime known (obviously).
Compile errors that report "expected pointer, found ..." are factored
out into convenience functions `checkPtrOperand` and `checkPtrType` and
have a note about function pointers.
Implemented `Value.hash` for functions, enum literals, and undefined values.
stage1 is not updated to this (yet?), so some workarounds and disabled
tests are needed to keep everything working. Should we update stage1 to
these new type semantics? Yes probably because I don't want to add too
much conditional compilation logic in the std lib for the different
backends.
2022-01-21 00:49:58 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-01-09 14:59:19 +00:00
|
|
|
|
stage2: type system treats fn ptr and body separately
This commit updates stage2 to enforce the property that the syntax
`fn()void` is a function *body* not a *pointer*. To get a pointer, the
syntax `*const fn()void` is required.
ZIR puts function alignment into the func instruction rather than the
decl because this way it makes it into function types. LLVM backend
respects function alignments.
Struct and Union have methods `fieldSrcLoc` to help look up source
locations of their fields. These trigger full loading, tokenization, and
parsing of source files, so should only be called once it is confirmed
that an error message needs to be printed.
There are some nice new error hints for explaining why a type is
required to be comptime, particularly for structs that contain function
body types.
`Type.requiresComptime` is now moved into Sema because it can fail and
might need to trigger field type resolution. Comptime pointer loading
takes into account types that do not have a well-defined memory layout
and does not try to compute a byte offset for them.
`fn()void` syntax no longer secretly makes a pointer. You get a function
body type, which requires comptime. However a pointer to a function body
can be runtime known (obviously).
Compile errors that report "expected pointer, found ..." are factored
out into convenience functions `checkPtrOperand` and `checkPtrType` and
have a note about function pointers.
Implemented `Value.hash` for functions, enum literals, and undefined values.
stage1 is not updated to this (yet?), so some workarounds and disabled
tests are needed to keep everything working. Should we update stage1 to
these new type semantics? Yes probably because I don't want to add too
much conditional compilation logic in the std lib for the different
backends.
2022-01-21 00:49:58 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2022-11-22 00:23:08 +02:00
|
|
|
|
stage2: type system treats fn ptr and body separately
This commit updates stage2 to enforce the property that the syntax
`fn()void` is a function *body* not a *pointer*. To get a pointer, the
syntax `*const fn()void` is required.
ZIR puts function alignment into the func instruction rather than the
decl because this way it makes it into function types. LLVM backend
respects function alignments.
Struct and Union have methods `fieldSrcLoc` to help look up source
locations of their fields. These trigger full loading, tokenization, and
parsing of source files, so should only be called once it is confirmed
that an error message needs to be printed.
There are some nice new error hints for explaining why a type is
required to be comptime, particularly for structs that contain function
body types.
`Type.requiresComptime` is now moved into Sema because it can fail and
might need to trigger field type resolution. Comptime pointer loading
takes into account types that do not have a well-defined memory layout
and does not try to compute a byte offset for them.
`fn()void` syntax no longer secretly makes a pointer. You get a function
body type, which requires comptime. However a pointer to a function body
can be runtime known (obviously).
Compile errors that report "expected pointer, found ..." are factored
out into convenience functions `checkPtrOperand` and `checkPtrType` and
have a note about function pointers.
Implemented `Value.hash` for functions, enum literals, and undefined values.
stage1 is not updated to this (yet?), so some workarounds and disabled
tests are needed to keep everything working. Should we update stage1 to
these new type semantics? Yes probably because I don't want to add too
much conditional compilation logic in the std lib for the different
backends.
2022-01-21 00:49:58 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2022-07-19 20:20:57 +03:00
|
|
|
|
|
|
|
|
|
2023-05-08 16:52:59 -07:00
|
|
|
|
2022-07-19 20:20:57 +03:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-05-08 16:52:59 -07:00
|
|
|
|
|
|
|
|
|
2022-07-19 20:20:57 +03:00
|
|
|
|
|
|
|
|
|
2023-05-08 16:52:59 -07:00
|
|
|
|
2022-07-19 20:20:57 +03:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-01-09 14:59:19 +00:00
|
|
|
|
|
|
|
|
|
2022-07-19 20:20:57 +03:00
|
|
|
|
2022-08-21 18:04:46 +03:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2022-07-19 20:20:57 +03:00
|
|
|
|
|
|
|
|
|
2023-06-22 18:46:56 +01:00
|
|
|
|
2022-07-19 20:20:57 +03:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2022-08-21 18:04:46 +03:00
|
|
|
|
2022-07-19 20:20:57 +03:00
|
|
|
|
|
|
|
|
|
2022-07-22 18:21:03 +03:00
|
|
|
|
2023-05-08 16:52:59 -07:00
|
|
|
|
2022-07-22 18:21:03 +03:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-05-08 16:52:59 -07:00
|
|
|
|
|
|
|
|
|
2022-07-22 18:21:03 +03:00
|
|
|
|
|
|
|
|
|
2023-05-08 16:52:59 -07:00
|
|
|
|
2022-07-22 18:21:03 +03:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-01-09 14:59:19 +00:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2022-07-22 18:21:03 +03:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-01-09 14:59:19 +00:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2022-07-22 18:21:03 +03:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-05-08 16:52:59 -07:00
|
|
|
|
2022-07-25 17:53:23 +03:00
|
|
|
|
2023-05-08 16:52:59 -07:00
|
|
|
|
|
|
|
|
|
2022-07-25 17:53:23 +03:00
|
|
|
|
|
|
|
|
|
2023-05-08 16:52:59 -07:00
|
|
|
|
2022-07-25 17:53:23 +03:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-01-09 14:59:19 +00:00
|
|
|
|
2022-07-25 17:53:23 +03:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-05-06 17:20:45 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2022-04-19 21:51:08 -07:00
|
|
|
|
2021-05-06 17:20:45 -07:00
|
|
|
|
2021-06-03 15:39:26 -05:00
|
|
|
|
2021-05-06 17:20:45 -07:00
|
|
|
|
2022-04-19 21:51:08 -07:00
|
|
|
|
|
|
|
|
|
2021-05-06 17:20:45 -07:00
|
|
|
|
2021-05-11 22:12:36 -07:00
|
|
|
|
|
|
|
|
|
2021-05-06 17:20:45 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2022-04-19 21:51:08 -07:00
|
|
|
|
|
|
|
|
|
stage2: fix deletion of Decls that get re-referenced
When scanDecls happens, we create stub Decl objects that
have not been semantically analyzed. When they get referenced,
they get semantically analyzed.
Before this commit, when they got unreferenced, they were completely
deleted, including deleted from the containing Namespace.
However, if the update did not cause the containing Namespace to get
deleted, for example, if `std.builtin.ExportOptions` is no longer
referenced, but `std.builtin` is still referenced, and then `ExportOptions`
gets referenced again, the Namespace would be incorrectly missing the
Decl, so we get an incorrect "no such member" error.
The solution is to, when dealing with a no longer referenced Decl
objects during an update, clear them to the state they would be in
on a fresh scanDecl, rather than completely deleting them.
2021-05-18 12:35:36 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-05-08 16:52:59 -07:00
|
|
|
|
compiler: eliminate Decl.value_arena and Sema.perm_arena
The main motivation for this commit is eliminating Decl.value_arena.
Everything else is dominoes.
Decl.name used to be stored in the GPA, now it is stored in InternPool.
It ended up being simpler to migrate other strings to be interned as
well, such as struct field names, union field names, and a few others.
This ended up requiring a big diff, sorry about that. But the changes
are pretty nice, we finally start to take advantage of InternPool's
existence.
global_error_set and error_name_list are simplified. Now it is a single
ArrayHashMap(NullTerminatedString, void) and the index is the error tag
value.
Module.tmp_hack_arena is re-introduced (it was removed in
eeff407941560ce8eb5b737b2436dfa93cfd3a0c) in order to deal with
comptime_args, optimized_order, and struct and union fields. After
structs and unions get moved into InternPool properly, tmp_hack_arena
can be deleted again.
2023-06-01 23:46:04 -07:00
|
|
|
|
2023-05-08 16:52:59 -07:00
|
|
|
|
|
|
|
|
|
stage2: fix deletion of Decls that get re-referenced
When scanDecls happens, we create stub Decl objects that
have not been semantically analyzed. When they get referenced,
they get semantically analyzed.
Before this commit, when they got unreferenced, they were completely
deleted, including deleted from the containing Namespace.
However, if the update did not cause the containing Namespace to get
deleted, for example, if `std.builtin.ExportOptions` is no longer
referenced, but `std.builtin` is still referenced, and then `ExportOptions`
gets referenced again, the Namespace would be incorrectly missing the
Decl, so we get an incorrect "no such member" error.
The solution is to, when dealing with a no longer referenced Decl
objects during an update, clear them to the state they would be in
on a fresh scanDecl, rather than completely deleting them.
2021-05-18 12:35:36 -07:00
|
|
|
|
2022-04-19 21:51:08 -07:00
|
|
|
|
|
|
|
|
|
2021-05-06 17:20:45 -07:00
|
|
|
|
2021-05-11 22:12:36 -07:00
|
|
|
|
2021-05-06 17:20:45 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-06-03 15:39:26 -05:00
|
|
|
|
|
|
|
|
|
2021-05-06 17:20:45 -07:00
|
|
|
|
|
|
|
|
|
2021-05-14 17:41:22 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
compiler: eliminate Decl.value_arena and Sema.perm_arena
The main motivation for this commit is eliminating Decl.value_arena.
Everything else is dominoes.
Decl.name used to be stored in the GPA, now it is stored in InternPool.
It ended up being simpler to migrate other strings to be interned as
well, such as struct field names, union field names, and a few others.
This ended up requiring a big diff, sorry about that. But the changes
are pretty nice, we finally start to take advantage of InternPool's
existence.
global_error_set and error_name_list are simplified. Now it is a single
ArrayHashMap(NullTerminatedString, void) and the index is the error tag
value.
Module.tmp_hack_arena is re-introduced (it was removed in
eeff407941560ce8eb5b737b2436dfa93cfd3a0c) in order to deal with
comptime_args, optimized_order, and struct and union fields. After
structs and unions get moved into InternPool properly, tmp_hack_arena
can be deleted again.
2023-06-01 23:46:04 -07:00
|
|
|
|
2021-05-14 17:41:22 -07:00
|
|
|
|
|
|
|
|
|
2021-06-03 15:39:26 -05:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2022-11-27 01:07:35 -07:00
|
|
|
|
2021-05-14 17:41:22 -07:00
|
|
|
|
2023-06-12 01:44:12 -04:00
|
|
|
|
2021-05-14 17:41:22 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2022-04-19 21:51:08 -07:00
|
|
|
|
2023-06-12 01:44:12 -04:00
|
|
|
|
|
|
|
|
|
2021-05-14 17:41:22 -07:00
|
|
|
|
|
|
|
|
|
2021-06-03 15:39:26 -05:00
|
|
|
|
2022-04-19 21:51:08 -07:00
|
|
|
|
2021-05-14 17:41:22 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-06-03 15:39:26 -05:00
|
|
|
|
2021-05-14 17:41:22 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2022-04-19 21:51:08 -07:00
|
|
|
|
2021-05-14 17:41:22 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
stage2: `zig test` now works with the LLVM backend
Frontend improvements:
* When compiling in `zig test` mode, put a task on the work queue to
analyze the main package root file. Normally, start code does
`_ = import("root");` to make Zig analyze the user's code, however in
the case of `zig test`, the root source file is the test runner.
Without this change, no tests are picked up.
* In the main pipeline, once semantic analysis is finished, if there
are no compile errors, populate the `test_functions` Decl with the
set of test functions picked up from semantic analysis.
* Value: add `array` and `slice` Tags.
LLVM backend improvements:
* Fix incremental updates of globals. Previously the
value of a global would not get replaced with a new value.
* Fix LLVM type of arrays. They were incorrectly sending
the ABI size as the element count.
* Remove the FuncGen parameter from genTypedValue. This function is for
generating global constants and there is no function available when
it is being called.
- The `ref_val` case is now commented out. I'd like to eliminate
`ref_val` as one of the possible Value Tags. Instead it should
always be done via `decl_ref`.
* Implement constant value generation for slices, arrays, and structs.
* Constant value generation for functions supports the `decl_ref` tag.
2021-07-27 14:06:42 -07:00
|
|
|
|
2022-11-15 16:50:20 +02:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
stage2: `zig test` now works with the LLVM backend
Frontend improvements:
* When compiling in `zig test` mode, put a task on the work queue to
analyze the main package root file. Normally, start code does
`_ = import("root");` to make Zig analyze the user's code, however in
the case of `zig test`, the root source file is the test runner.
Without this change, no tests are picked up.
* In the main pipeline, once semantic analysis is finished, if there
are no compile errors, populate the `test_functions` Decl with the
set of test functions picked up from semantic analysis.
* Value: add `array` and `slice` Tags.
LLVM backend improvements:
* Fix incremental updates of globals. Previously the
value of a global would not get replaced with a new value.
* Fix LLVM type of arrays. They were incorrectly sending
the ABI size as the element count.
* Remove the FuncGen parameter from genTypedValue. This function is for
generating global constants and there is no function available when
it is being called.
- The `ref_val` case is now commented out. I'd like to eliminate
`ref_val` as one of the possible Value Tags. Instead it should
always be done via `decl_ref`.
* Implement constant value generation for slices, arrays, and structs.
* Constant value generation for functions supports the `decl_ref` tag.
2021-07-27 14:06:42 -07:00
|
|
|
|
2023-06-02 00:53:33 -07:00
|
|
|
|
2023-10-06 21:29:08 -07:00
|
|
|
|
2023-10-06 17:05:58 -07:00
|
|
|
|
2022-04-19 21:51:08 -07:00
|
|
|
|
2023-05-08 16:52:59 -07:00
|
|
|
|
2023-06-02 00:53:33 -07:00
|
|
|
|
2023-05-08 16:52:59 -07:00
|
|
|
|
compiler: eliminate Decl.value_arena and Sema.perm_arena
The main motivation for this commit is eliminating Decl.value_arena.
Everything else is dominoes.
Decl.name used to be stored in the GPA, now it is stored in InternPool.
It ended up being simpler to migrate other strings to be interned as
well, such as struct field names, union field names, and a few others.
This ended up requiring a big diff, sorry about that. But the changes
are pretty nice, we finally start to take advantage of InternPool's
existence.
global_error_set and error_name_list are simplified. Now it is a single
ArrayHashMap(NullTerminatedString, void) and the index is the error tag
value.
Module.tmp_hack_arena is re-introduced (it was removed in
eeff407941560ce8eb5b737b2436dfa93cfd3a0c) in order to deal with
comptime_args, optimized_order, and struct and union fields. After
structs and unions get moved into InternPool properly, tmp_hack_arena
can be deleted again.
2023-06-01 23:46:04 -07:00
|
|
|
|
2023-05-08 16:52:59 -07:00
|
|
|
|
|
|
|
|
|
2022-11-15 16:50:20 +02:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2022-04-19 21:51:08 -07:00
|
|
|
|
2023-05-22 07:58:02 -04:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
stage2: `zig test` now works with the LLVM backend
Frontend improvements:
* When compiling in `zig test` mode, put a task on the work queue to
analyze the main package root file. Normally, start code does
`_ = import("root");` to make Zig analyze the user's code, however in
the case of `zig test`, the root source file is the test runner.
Without this change, no tests are picked up.
* In the main pipeline, once semantic analysis is finished, if there
are no compile errors, populate the `test_functions` Decl with the
set of test functions picked up from semantic analysis.
* Value: add `array` and `slice` Tags.
LLVM backend improvements:
* Fix incremental updates of globals. Previously the
value of a global would not get replaced with a new value.
* Fix LLVM type of arrays. They were incorrectly sending
the ABI size as the element count.
* Remove the FuncGen parameter from genTypedValue. This function is for
generating global constants and there is no function available when
it is being called.
- The `ref_val` case is now commented out. I'd like to eliminate
`ref_val` as one of the possible Value Tags. Instead it should
always be done via `decl_ref`.
* Implement constant value generation for slices, arrays, and structs.
* Constant value generation for functions supports the `decl_ref` tag.
2021-07-27 14:06:42 -07:00
|
|
|
|
2022-04-19 21:51:08 -07:00
|
|
|
|
stage2: `zig test` now works with the LLVM backend
Frontend improvements:
* When compiling in `zig test` mode, put a task on the work queue to
analyze the main package root file. Normally, start code does
`_ = import("root");` to make Zig analyze the user's code, however in
the case of `zig test`, the root source file is the test runner.
Without this change, no tests are picked up.
* In the main pipeline, once semantic analysis is finished, if there
are no compile errors, populate the `test_functions` Decl with the
set of test functions picked up from semantic analysis.
* Value: add `array` and `slice` Tags.
LLVM backend improvements:
* Fix incremental updates of globals. Previously the
value of a global would not get replaced with a new value.
* Fix LLVM type of arrays. They were incorrectly sending
the ABI size as the element count.
* Remove the FuncGen parameter from genTypedValue. This function is for
generating global constants and there is no function available when
it is being called.
- The `ref_val` case is now commented out. I'd like to eliminate
`ref_val` as one of the possible Value Tags. Instead it should
always be done via `decl_ref`.
* Implement constant value generation for slices, arrays, and structs.
* Constant value generation for functions supports the `decl_ref` tag.
2021-07-27 14:06:42 -07:00
|
|
|
|
|
|
|
|
|
2023-05-25 05:47:25 -04:00
|
|
|
|
|
|
|
|
|
2021-10-28 15:59:14 -07:00
|
|
|
|
|
|
|
|
|
2023-05-25 05:47:25 -04:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-10-28 15:59:14 -07:00
|
|
|
|
2023-05-25 05:47:25 -04:00
|
|
|
|
2022-04-19 21:51:08 -07:00
|
|
|
|
2023-06-02 18:49:40 -04:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2022-04-19 21:51:08 -07:00
|
|
|
|
2023-05-25 05:47:25 -04:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-09-14 19:59:58 -07:00
|
|
|
|
2023-05-25 05:47:25 -04:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
stage2: `zig test` now works with the LLVM backend
Frontend improvements:
* When compiling in `zig test` mode, put a task on the work queue to
analyze the main package root file. Normally, start code does
`_ = import("root");` to make Zig analyze the user's code, however in
the case of `zig test`, the root source file is the test runner.
Without this change, no tests are picked up.
* In the main pipeline, once semantic analysis is finished, if there
are no compile errors, populate the `test_functions` Decl with the
set of test functions picked up from semantic analysis.
* Value: add `array` and `slice` Tags.
LLVM backend improvements:
* Fix incremental updates of globals. Previously the
value of a global would not get replaced with a new value.
* Fix LLVM type of arrays. They were incorrectly sending
the ABI size as the element count.
* Remove the FuncGen parameter from genTypedValue. This function is for
generating global constants and there is no function available when
it is being called.
- The `ref_val` case is now commented out. I'd like to eliminate
`ref_val` as one of the possible Value Tags. Instead it should
always be done via `decl_ref`.
* Implement constant value generation for slices, arrays, and structs.
* Constant value generation for functions supports the `decl_ref` tag.
2021-07-27 14:06:42 -07:00
|
|
|
|
2022-04-19 21:51:08 -07:00
|
|
|
|
stage2: `zig test` now works with the LLVM backend
Frontend improvements:
* When compiling in `zig test` mode, put a task on the work queue to
analyze the main package root file. Normally, start code does
`_ = import("root");` to make Zig analyze the user's code, however in
the case of `zig test`, the root source file is the test runner.
Without this change, no tests are picked up.
* In the main pipeline, once semantic analysis is finished, if there
are no compile errors, populate the `test_functions` Decl with the
set of test functions picked up from semantic analysis.
* Value: add `array` and `slice` Tags.
LLVM backend improvements:
* Fix incremental updates of globals. Previously the
value of a global would not get replaced with a new value.
* Fix LLVM type of arrays. They were incorrectly sending
the ABI size as the element count.
* Remove the FuncGen parameter from genTypedValue. This function is for
generating global constants and there is no function available when
it is being called.
- The `ref_val` case is now commented out. I'd like to eliminate
`ref_val` as one of the possible Value Tags. Instead it should
always be done via `decl_ref`.
* Implement constant value generation for slices, arrays, and structs.
* Constant value generation for functions supports the `decl_ref` tag.
2021-07-27 14:06:42 -07:00
|
|
|
|
2023-05-25 05:47:25 -04:00
|
|
|
|
|
|
|
|
|
2022-04-19 21:51:08 -07:00
|
|
|
|
stage2: `zig test` now works with the LLVM backend
Frontend improvements:
* When compiling in `zig test` mode, put a task on the work queue to
analyze the main package root file. Normally, start code does
`_ = import("root");` to make Zig analyze the user's code, however in
the case of `zig test`, the root source file is the test runner.
Without this change, no tests are picked up.
* In the main pipeline, once semantic analysis is finished, if there
are no compile errors, populate the `test_functions` Decl with the
set of test functions picked up from semantic analysis.
* Value: add `array` and `slice` Tags.
LLVM backend improvements:
* Fix incremental updates of globals. Previously the
value of a global would not get replaced with a new value.
* Fix LLVM type of arrays. They were incorrectly sending
the ABI size as the element count.
* Remove the FuncGen parameter from genTypedValue. This function is for
generating global constants and there is no function available when
it is being called.
- The `ref_val` case is now commented out. I'd like to eliminate
`ref_val` as one of the possible Value Tags. Instead it should
always be done via `decl_ref`.
* Implement constant value generation for slices, arrays, and structs.
* Constant value generation for functions supports the `decl_ref` tag.
2021-07-27 14:06:42 -07:00
|
|
|
|
2023-05-22 07:58:02 -04:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-05-26 16:41:17 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-05-22 07:58:02 -04:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-05-26 16:41:17 -07:00
|
|
|
|
2023-05-30 20:23:51 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-05-26 16:41:17 -07:00
|
|
|
|
2023-05-22 07:58:02 -04:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
stage2: `zig test` now works with the LLVM backend
Frontend improvements:
* When compiling in `zig test` mode, put a task on the work queue to
analyze the main package root file. Normally, start code does
`_ = import("root");` to make Zig analyze the user's code, however in
the case of `zig test`, the root source file is the test runner.
Without this change, no tests are picked up.
* In the main pipeline, once semantic analysis is finished, if there
are no compile errors, populate the `test_functions` Decl with the
set of test functions picked up from semantic analysis.
* Value: add `array` and `slice` Tags.
LLVM backend improvements:
* Fix incremental updates of globals. Previously the
value of a global would not get replaced with a new value.
* Fix LLVM type of arrays. They were incorrectly sending
the ABI size as the element count.
* Remove the FuncGen parameter from genTypedValue. This function is for
generating global constants and there is no function available when
it is being called.
- The `ref_val` case is now commented out. I'd like to eliminate
`ref_val` as one of the possible Value Tags. Instead it should
always be done via `decl_ref`.
* Implement constant value generation for slices, arrays, and structs.
* Constant value generation for functions supports the `decl_ref` tag.
2021-07-27 14:06:42 -07:00
|
|
|
|
2023-05-25 05:47:25 -04:00
|
|
|
|
|
|
|
|
|
2023-05-22 07:58:02 -04:00
|
|
|
|
2023-05-25 05:47:25 -04:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-09-14 19:59:58 -07:00
|
|
|
|
2023-05-25 05:47:25 -04:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
stage2: `zig test` now works with the LLVM backend
Frontend improvements:
* When compiling in `zig test` mode, put a task on the work queue to
analyze the main package root file. Normally, start code does
`_ = import("root");` to make Zig analyze the user's code, however in
the case of `zig test`, the root source file is the test runner.
Without this change, no tests are picked up.
* In the main pipeline, once semantic analysis is finished, if there
are no compile errors, populate the `test_functions` Decl with the
set of test functions picked up from semantic analysis.
* Value: add `array` and `slice` Tags.
LLVM backend improvements:
* Fix incremental updates of globals. Previously the
value of a global would not get replaced with a new value.
* Fix LLVM type of arrays. They were incorrectly sending
the ABI size as the element count.
* Remove the FuncGen parameter from genTypedValue. This function is for
generating global constants and there is no function available when
it is being called.
- The `ref_val` case is now commented out. I'd like to eliminate
`ref_val` as one of the possible Value Tags. Instead it should
always be done via `decl_ref`.
* Implement constant value generation for slices, arrays, and structs.
* Constant value generation for functions supports the `decl_ref` tag.
2021-07-27 14:06:42 -07:00
|
|
|
|
|
|
|
|
|
2022-04-19 21:51:08 -07:00
|
|
|
|
stage2: `zig test` now works with the LLVM backend
Frontend improvements:
* When compiling in `zig test` mode, put a task on the work queue to
analyze the main package root file. Normally, start code does
`_ = import("root");` to make Zig analyze the user's code, however in
the case of `zig test`, the root source file is the test runner.
Without this change, no tests are picked up.
* In the main pipeline, once semantic analysis is finished, if there
are no compile errors, populate the `test_functions` Decl with the
set of test functions picked up from semantic analysis.
* Value: add `array` and `slice` Tags.
LLVM backend improvements:
* Fix incremental updates of globals. Previously the
value of a global would not get replaced with a new value.
* Fix LLVM type of arrays. They were incorrectly sending
the ABI size as the element count.
* Remove the FuncGen parameter from genTypedValue. This function is for
generating global constants and there is no function available when
it is being called.
- The `ref_val` case is now commented out. I'd like to eliminate
`ref_val` as one of the possible Value Tags. Instead it should
always be done via `decl_ref`.
* Implement constant value generation for slices, arrays, and structs.
* Constant value generation for functions supports the `decl_ref` tag.
2021-07-27 14:06:42 -07:00
|
|
|
|
2022-04-19 21:51:08 -07:00
|
|
|
|
stage2: `zig test` now works with the LLVM backend
Frontend improvements:
* When compiling in `zig test` mode, put a task on the work queue to
analyze the main package root file. Normally, start code does
`_ = import("root");` to make Zig analyze the user's code, however in
the case of `zig test`, the root source file is the test runner.
Without this change, no tests are picked up.
* In the main pipeline, once semantic analysis is finished, if there
are no compile errors, populate the `test_functions` Decl with the
set of test functions picked up from semantic analysis.
* Value: add `array` and `slice` Tags.
LLVM backend improvements:
* Fix incremental updates of globals. Previously the
value of a global would not get replaced with a new value.
* Fix LLVM type of arrays. They were incorrectly sending
the ABI size as the element count.
* Remove the FuncGen parameter from genTypedValue. This function is for
generating global constants and there is no function available when
it is being called.
- The `ref_val` case is now commented out. I'd like to eliminate
`ref_val` as one of the possible Value Tags. Instead it should
always be done via `decl_ref`.
* Implement constant value generation for slices, arrays, and structs.
* Constant value generation for functions supports the `decl_ref` tag.
2021-07-27 14:06:42 -07:00
|
|
|
|
|
|
|
|
|
2023-05-22 07:58:02 -04:00
|
|
|
|
2023-05-30 20:23:51 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-05-22 07:58:02 -04:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-05-25 05:47:25 -04:00
|
|
|
|
2023-05-22 07:58:02 -04:00
|
|
|
|
2023-05-25 05:47:25 -04:00
|
|
|
|
2023-05-22 07:58:02 -04:00
|
|
|
|
2023-06-02 00:53:33 -07:00
|
|
|
|
stage2: `zig test` now works with the LLVM backend
Frontend improvements:
* When compiling in `zig test` mode, put a task on the work queue to
analyze the main package root file. Normally, start code does
`_ = import("root");` to make Zig analyze the user's code, however in
the case of `zig test`, the root source file is the test runner.
Without this change, no tests are picked up.
* In the main pipeline, once semantic analysis is finished, if there
are no compile errors, populate the `test_functions` Decl with the
set of test functions picked up from semantic analysis.
* Value: add `array` and `slice` Tags.
LLVM backend improvements:
* Fix incremental updates of globals. Previously the
value of a global would not get replaced with a new value.
* Fix LLVM type of arrays. They were incorrectly sending
the ABI size as the element count.
* Remove the FuncGen parameter from genTypedValue. This function is for
generating global constants and there is no function available when
it is being called.
- The `ref_val` case is now commented out. I'd like to eliminate
`ref_val` as one of the possible Value Tags. Instead it should
always be done via `decl_ref`.
* Implement constant value generation for slices, arrays, and structs.
* Constant value generation for functions supports the `decl_ref` tag.
2021-07-27 14:06:42 -07:00
|
|
|
|
2023-05-22 07:58:02 -04:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
stage2: `zig test` now works with the LLVM backend
Frontend improvements:
* When compiling in `zig test` mode, put a task on the work queue to
analyze the main package root file. Normally, start code does
`_ = import("root");` to make Zig analyze the user's code, however in
the case of `zig test`, the root source file is the test runner.
Without this change, no tests are picked up.
* In the main pipeline, once semantic analysis is finished, if there
are no compile errors, populate the `test_functions` Decl with the
set of test functions picked up from semantic analysis.
* Value: add `array` and `slice` Tags.
LLVM backend improvements:
* Fix incremental updates of globals. Previously the
value of a global would not get replaced with a new value.
* Fix LLVM type of arrays. They were incorrectly sending
the ABI size as the element count.
* Remove the FuncGen parameter from genTypedValue. This function is for
generating global constants and there is no function available when
it is being called.
- The `ref_val` case is now commented out. I'd like to eliminate
`ref_val` as one of the possible Value Tags. Instead it should
always be done via `decl_ref`.
* Implement constant value generation for slices, arrays, and structs.
* Constant value generation for functions supports the `decl_ref` tag.
2021-07-27 14:06:42 -07:00
|
|
|
|
2022-04-19 21:51:08 -07:00
|
|
|
|
stage2: `zig test` now works with the LLVM backend
Frontend improvements:
* When compiling in `zig test` mode, put a task on the work queue to
analyze the main package root file. Normally, start code does
`_ = import("root");` to make Zig analyze the user's code, however in
the case of `zig test`, the root source file is the test runner.
Without this change, no tests are picked up.
* In the main pipeline, once semantic analysis is finished, if there
are no compile errors, populate the `test_functions` Decl with the
set of test functions picked up from semantic analysis.
* Value: add `array` and `slice` Tags.
LLVM backend improvements:
* Fix incremental updates of globals. Previously the
value of a global would not get replaced with a new value.
* Fix LLVM type of arrays. They were incorrectly sending
the ABI size as the element count.
* Remove the FuncGen parameter from genTypedValue. This function is for
generating global constants and there is no function available when
it is being called.
- The `ref_val` case is now commented out. I'd like to eliminate
`ref_val` as one of the possible Value Tags. Instead it should
always be done via `decl_ref`.
* Implement constant value generation for slices, arrays, and structs.
* Constant value generation for functions supports the `decl_ref` tag.
2021-07-27 14:06:42 -07:00
|
|
|
|
|
|
|
|
|
2022-04-19 21:51:08 -07:00
|
|
|
|
2022-02-17 13:12:31 -07:00
|
|
|
|
|
|
|
|
|
2022-10-15 02:34:34 -04:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-03-16 01:32:43 -04:00
|
|
|
|
2022-10-15 02:34:34 -04:00
|
|
|
|
|
|
|
|
|
2022-02-17 13:12:31 -07:00
|
|
|
|
2022-04-19 21:51:08 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
stage2: `zig test` now works with the LLVM backend
Frontend improvements:
* When compiling in `zig test` mode, put a task on the work queue to
analyze the main package root file. Normally, start code does
`_ = import("root");` to make Zig analyze the user's code, however in
the case of `zig test`, the root source file is the test runner.
Without this change, no tests are picked up.
* In the main pipeline, once semantic analysis is finished, if there
are no compile errors, populate the `test_functions` Decl with the
set of test functions picked up from semantic analysis.
* Value: add `array` and `slice` Tags.
LLVM backend improvements:
* Fix incremental updates of globals. Previously the
value of a global would not get replaced with a new value.
* Fix LLVM type of arrays. They were incorrectly sending
the ABI size as the element count.
* Remove the FuncGen parameter from genTypedValue. This function is for
generating global constants and there is no function available when
it is being called.
- The `ref_val` case is now commented out. I'd like to eliminate
`ref_val` as one of the possible Value Tags. Instead it should
always be done via `decl_ref`.
* Implement constant value generation for slices, arrays, and structs.
* Constant value generation for functions supports the `decl_ref` tag.
2021-07-27 14:06:42 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2022-04-19 21:51:08 -07:00
|
|
|
|
stage2: `zig test` now works with the LLVM backend
Frontend improvements:
* When compiling in `zig test` mode, put a task on the work queue to
analyze the main package root file. Normally, start code does
`_ = import("root");` to make Zig analyze the user's code, however in
the case of `zig test`, the root source file is the test runner.
Without this change, no tests are picked up.
* In the main pipeline, once semantic analysis is finished, if there
are no compile errors, populate the `test_functions` Decl with the
set of test functions picked up from semantic analysis.
* Value: add `array` and `slice` Tags.
LLVM backend improvements:
* Fix incremental updates of globals. Previously the
value of a global would not get replaced with a new value.
* Fix LLVM type of arrays. They were incorrectly sending
the ABI size as the element count.
* Remove the FuncGen parameter from genTypedValue. This function is for
generating global constants and there is no function available when
it is being called.
- The `ref_val` case is now commented out. I'd like to eliminate
`ref_val` as one of the possible Value Tags. Instead it should
always be done via `decl_ref`.
* Implement constant value generation for slices, arrays, and structs.
* Constant value generation for functions supports the `decl_ref` tag.
2021-07-27 14:06:42 -07:00
|
|
|
|
2023-05-08 16:52:59 -07:00
|
|
|
|
stage2: `zig test` now works with the LLVM backend
Frontend improvements:
* When compiling in `zig test` mode, put a task on the work queue to
analyze the main package root file. Normally, start code does
`_ = import("root");` to make Zig analyze the user's code, however in
the case of `zig test`, the root source file is the test runner.
Without this change, no tests are picked up.
* In the main pipeline, once semantic analysis is finished, if there
are no compile errors, populate the `test_functions` Decl with the
set of test functions picked up from semantic analysis.
* Value: add `array` and `slice` Tags.
LLVM backend improvements:
* Fix incremental updates of globals. Previously the
value of a global would not get replaced with a new value.
* Fix LLVM type of arrays. They were incorrectly sending
the ABI size as the element count.
* Remove the FuncGen parameter from genTypedValue. This function is for
generating global constants and there is no function available when
it is being called.
- The `ref_val` case is now commented out. I'd like to eliminate
`ref_val` as one of the possible Value Tags. Instead it should
always be done via `decl_ref`.
* Implement constant value generation for slices, arrays, and structs.
* Constant value generation for functions supports the `decl_ref` tag.
2021-07-27 14:06:42 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-12-30 20:49:02 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2022-04-19 21:51:08 -07:00
|
|
|
|
2023-05-26 00:24:29 -04:00
|
|
|
|
2023-05-25 05:47:25 -04:00
|
|
|
|
2023-05-26 00:24:29 -04:00
|
|
|
|
|
|
|
|
|
rework generic function calls
Abridged summary:
* Move `Module.Fn` into `InternPool`.
* Delete a lot of confusing and problematic `Sema` logic related to
generic function calls.
This commit removes `Module.Fn` and replaces it with two new
`InternPool.Tag` values:
* `func_decl` - corresponding to a function declared in the source
code. This one contains line/column numbers, zir_body_inst, etc.
* `func_instance` - one for each monomorphization of a generic
function. Contains a reference to the `func_decl` from whence the
instantiation came, along with the `comptime` parameter values (or
types in the case of `anytype`)
Since `InternPool` provides deduplication on these values, these fields
are now deleted from `Module`:
* `monomorphed_func_keys`
* `monomorphed_funcs`
* `align_stack_fns`
Instead of these, Sema logic for generic function instantiation now
unconditionally evaluates the function prototype expression for every
generic callsite. This is technically required in order for type
coercions to work. The previous code had some dubious, probably wrong
hacks to make things work, such as `hashUncoerced`. I'm not 100% sure
how we were able to eliminate that function and still pass all the
behavior tests, but I'm pretty sure things were still broken without
doing type coercion for every generic function call argument.
After the function prototype is evaluated, it produces a deduplicated
`func_instance` `InternPool.Index` which can then be used for the
generic function call.
Some other nice things made by this simplification are the removal of
`comptime_args_fn_inst` and `preallocated_new_func` from `Sema`, and the
messy logic associated with them.
I have not yet been able to measure the perf of this against master
branch. On one hand, it reduces memory usage and pointer chasing of the
most heavily used `InternPool` Tag - function bodies - but on the other
hand, it does evaluate function prototype expressions more than before.
We will soon find out.
2023-07-03 22:09:30 -07:00
|
|
|
|
2023-05-25 05:47:25 -04:00
|
|
|
|
|
|
|
|
|
2023-05-26 00:24:29 -04:00
|
|
|
|
2022-04-19 21:51:08 -07:00
|
|
|
|
2023-05-25 05:47:25 -04:00
|
|
|
|
|
|
|
|
|
2023-05-26 00:24:29 -04:00
|
|
|
|
2023-09-24 19:22:25 -07:00
|
|
|
|
2023-05-26 00:24:29 -04:00
|
|
|
|
2023-05-25 05:47:25 -04:00
|
|
|
|
2023-05-26 00:24:29 -04:00
|
|
|
|
|
|
|
|
|
2023-05-25 05:47:25 -04:00
|
|
|
|
2023-05-26 00:24:29 -04:00
|
|
|
|
2022-04-19 21:51:08 -07:00
|
|
|
|
2023-05-26 00:24:29 -04:00
|
|
|
|
2023-05-25 05:47:25 -04:00
|
|
|
|
2023-05-26 00:24:29 -04:00
|
|
|
|
2023-05-25 05:47:25 -04:00
|
|
|
|
2023-09-23 13:03:03 -04:00
|
|
|
|
2023-05-26 00:24:29 -04:00
|
|
|
|
2023-05-25 05:47:25 -04:00
|
|
|
|
|
|
|
|
|
2022-04-19 21:51:08 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-05-26 00:24:29 -04:00
|
|
|
|
2022-04-19 21:51:08 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-07-10 10:52:17 -04:00
|
|
|
|
2023-05-26 00:24:29 -04:00
|
|
|
|
2022-04-19 21:51:08 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-05-26 00:24:29 -04:00
|
|
|
|
2022-04-19 21:51:08 -07:00
|
|
|
|
|
|
|
|
|
2023-05-26 00:24:29 -04:00
|
|
|
|
2022-04-19 21:51:08 -07:00
|
|
|
|
|
|
|
|
|
2022-05-04 20:38:53 -07:00
|
|
|
|
|
|
|
|
|
2023-06-19 01:18:40 -04:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2022-05-04 20:38:53 -07:00
|
|
|
|
2022-08-18 10:49:09 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2022-11-27 01:07:35 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-01-05 13:38:14 +02:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-01-16 19:46:41 +02:00
|
|
|
|
2023-06-24 15:03:15 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-01-05 13:38:14 +02:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-04-24 19:14:54 -07:00
|
|
|
|
|
|
|
|
|
2023-01-05 13:38:14 +02:00
|
|
|
|
|
|
|
|
|
2023-04-16 15:10:45 +02:00
|
|
|
|
2023-01-16 19:46:41 +02:00
|
|
|
|
2023-06-24 15:03:15 -07:00
|
|
|
|
2023-01-05 13:38:14 +02:00
|
|
|
|
|
|
|
|
|
2023-05-02 15:01:45 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-05-26 03:41:35 -04:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-05-02 15:01:45 -07:00
|
|
|
|
2023-05-26 14:58:07 -04:00
|
|
|
|
2023-05-02 15:01:45 -07:00
|
|
|
|
|
|
|
|
|
2023-05-26 14:58:07 -04:00
|
|
|
|
2023-05-02 15:01:45 -07:00
|
|
|
|
|
|
|
|
|
2023-05-04 20:30:25 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-05-26 14:58:07 -04:00
|
|
|
|
|
|
|
|
|
2023-05-30 20:23:51 -07:00
|
|
|
|
2023-05-29 22:22:05 -04:00
|
|
|
|
2023-05-26 14:58:07 -04:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-09-20 17:57:18 -07:00
|
|
|
|
2023-08-24 20:43:43 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-05-07 15:23:12 +01:00
|
|
|
|
2023-05-26 14:58:07 -04:00
|
|
|
|
2023-05-30 20:23:51 -07:00
|
|
|
|
2023-05-26 21:22:34 -04:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-09-20 17:57:18 -07:00
|
|
|
|
2023-05-30 20:23:51 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-05-26 21:22:34 -04:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-06-15 13:14:16 +06:00
|
|
|
|
2023-05-26 14:58:07 -04:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-05-04 20:30:25 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-05-30 20:23:51 -07:00
|
|
|
|
2023-05-04 20:30:25 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-05-30 20:23:51 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-05-04 20:30:25 -07:00
|
|
|
|
|
|
|
|
|
2023-05-29 07:30:30 -04:00
|
|
|
|
2023-05-30 20:23:51 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-05-29 07:30:30 -04:00
|
|
|
|
|
|
|
|
|
2023-05-18 22:02:55 -07:00
|
|
|
|
2023-06-20 14:02:04 -04:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-05-18 22:02:55 -07:00
|
|
|
|
|
|
|
|
|
rework generic function calls
Abridged summary:
* Move `Module.Fn` into `InternPool`.
* Delete a lot of confusing and problematic `Sema` logic related to
generic function calls.
This commit removes `Module.Fn` and replaces it with two new
`InternPool.Tag` values:
* `func_decl` - corresponding to a function declared in the source
code. This one contains line/column numbers, zir_body_inst, etc.
* `func_instance` - one for each monomorphization of a generic
function. Contains a reference to the `func_decl` from whence the
instantiation came, along with the `comptime` parameter values (or
types in the case of `anytype`)
Since `InternPool` provides deduplication on these values, these fields
are now deleted from `Module`:
* `monomorphed_func_keys`
* `monomorphed_funcs`
* `align_stack_fns`
Instead of these, Sema logic for generic function instantiation now
unconditionally evaluates the function prototype expression for every
generic callsite. This is technically required in order for type
coercions to work. The previous code had some dubious, probably wrong
hacks to make things work, such as `hashUncoerced`. I'm not 100% sure
how we were able to eliminate that function and still pass all the
behavior tests, but I'm pretty sure things were still broken without
doing type coercion for every generic function call argument.
After the function prototype is evaluated, it produces a deduplicated
`func_instance` `InternPool.Index` which can then be used for the
generic function call.
Some other nice things made by this simplification are the removal of
`comptime_args_fn_inst` and `preallocated_new_func` from `Sema`, and the
messy logic associated with them.
I have not yet been able to measure the perf of this against master
branch. On one hand, it reduces memory usage and pointer chasing of the
most heavily used `InternPool` Tag - function bodies - but on the other
hand, it does evaluate function prototype expressions more than before.
We will soon find out.
2023-07-03 22:09:30 -07:00
|
|
|
|
|
|
|
|
|
2023-05-15 20:09:54 -07:00
|
|
|
|
|
|
|
|
|
2023-05-18 19:05:46 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-05-20 12:09:07 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-06-12 01:44:12 -04:00
|
|
|
|
|
|
|
|
|
2023-07-08 23:39:37 -07:00
|
|
|
|
|
|
|
|
|
2023-05-20 12:09:07 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-07-08 23:39:37 -07:00
|
|
|
|
2023-05-20 12:09:07 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-06-14 18:10:47 +01:00
|
|
|
|
2023-05-08 11:51:32 -07:00
|
|
|
|
2023-08-27 17:14:31 -04:00
|
|
|
|
2023-05-08 11:51:32 -07:00
|
|
|
|
2023-05-25 05:47:25 -04:00
|
|
|
|
2023-06-14 18:10:47 +01:00
|
|
|
|
2023-05-08 11:51:32 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-05-12 16:22:37 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-05-25 05:47:25 -04:00
|
|
|
|
2023-05-12 16:22:37 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-05-25 05:47:25 -04:00
|
|
|
|
2023-05-12 16:22:37 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-05-25 05:47:25 -04:00
|
|
|
|
2023-05-12 16:22:37 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-05-25 05:47:25 -04:00
|
|
|
|
2023-08-17 14:25:18 -07:00
|
|
|
|
2023-05-12 16:22:37 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-08-11 14:02:52 +01:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-05-06 19:20:52 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-08-11 14:02:52 +01:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-05-06 19:20:52 -07:00
|
|
|
|
|
|
|
|
|
2023-05-25 05:47:25 -04:00
|
|
|
|
2023-05-06 19:20:52 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-05-25 05:47:25 -04:00
|
|
|
|
2023-05-06 19:20:52 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-05-25 05:47:25 -04:00
|
|
|
|
2023-05-06 19:20:52 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-05-10 17:21:22 -07:00
|
|
|
|
|
|
|
|
|
2023-05-25 05:47:25 -04:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-05-10 17:21:22 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-05-09 17:06:10 +01:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-06-22 18:46:56 +01:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-05-09 17:06:10 +01:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-05-25 05:47:25 -04:00
|
|
|
|
2023-05-09 17:06:10 +01:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-05-15 20:09:54 -07:00
|
|
|
|
|
|
|
|
|
2023-05-25 05:47:25 -04:00
|
|
|
|
2023-05-15 20:09:54 -07:00
|
|
|
|
2023-05-25 05:47:25 -04:00
|
|
|
|
2023-05-15 20:09:54 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-05-02 15:01:45 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-05-12 16:22:37 -07:00
|
|
|
|
|
|
|
|
|
2023-05-02 15:01:45 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-05-12 16:22:37 -07:00
|
|
|
|
2023-05-06 19:20:52 -07:00
|
|
|
|
2023-05-25 05:47:25 -04:00
|
|
|
|
2023-05-06 19:20:52 -07:00
|
|
|
|
|
|
|
|
|
2023-06-15 13:14:16 +06:00
|
|
|
|
2023-05-02 15:01:45 -07:00
|
|
|
|
2023-05-06 19:20:52 -07:00
|
|
|
|
|
|
|
|
|
2023-06-22 18:46:56 +01:00
|
|
|
|
2023-05-02 15:01:45 -07:00
|
|
|
|
2023-05-06 19:20:52 -07:00
|
|
|
|
2023-06-15 13:14:16 +06:00
|
|
|
|
2023-05-02 15:01:45 -07:00
|
|
|
|
2023-05-06 19:20:52 -07:00
|
|
|
|
2023-06-22 18:46:56 +01:00
|
|
|
|
2023-05-06 19:20:52 -07:00
|
|
|
|
|
|
|
|
|
2023-07-02 11:40:01 -04:00
|
|
|
|
2023-05-06 19:20:52 -07:00
|
|
|
|
2023-06-22 18:46:56 +01:00
|
|
|
|
2023-05-06 19:20:52 -07:00
|
|
|
|
2023-05-22 07:58:02 -04:00
|
|
|
|
2023-08-24 20:43:43 -07:00
|
|
|
|
2023-05-22 07:58:02 -04:00
|
|
|
|
|
|
|
|
|
2023-06-15 13:14:16 +06:00
|
|
|
|
2023-05-22 07:58:02 -04:00
|
|
|
|
2023-05-02 15:01:45 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-05-07 15:38:31 -07:00
|
|
|
|
2023-05-02 15:01:45 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-05-07 15:38:31 -07:00
|
|
|
|
2023-05-02 15:01:45 -07:00
|
|
|
|
|
|
|
|
|
2023-06-20 14:02:04 -04:00
|
|
|
|
2023-05-02 15:01:45 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-05-28 10:33:59 -04:00
|
|
|
|
2023-05-02 15:01:45 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-06-20 14:02:04 -04:00
|
|
|
|
2023-05-02 15:01:45 -07:00
|
|
|
|
2023-06-20 14:02:04 -04:00
|
|
|
|
2023-05-02 15:01:45 -07:00
|
|
|
|
2023-06-20 14:02:04 -04:00
|
|
|
|
2023-05-02 15:01:45 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-06-20 14:02:04 -04:00
|
|
|
|
2023-05-02 15:01:45 -07:00
|
|
|
|
2023-05-08 16:52:59 -07:00
|
|
|
|
|
|
|
|
|
2023-05-10 12:16:24 -07:00
|
|
|
|
2023-05-08 16:52:59 -07:00
|
|
|
|
|
|
|
|
|
compiler: eliminate Decl.value_arena and Sema.perm_arena
The main motivation for this commit is eliminating Decl.value_arena.
Everything else is dominoes.
Decl.name used to be stored in the GPA, now it is stored in InternPool.
It ended up being simpler to migrate other strings to be interned as
well, such as struct field names, union field names, and a few others.
This ended up requiring a big diff, sorry about that. But the changes
are pretty nice, we finally start to take advantage of InternPool's
existence.
global_error_set and error_name_list are simplified. Now it is a single
ArrayHashMap(NullTerminatedString, void) and the index is the error tag
value.
Module.tmp_hack_arena is re-introduced (it was removed in
eeff407941560ce8eb5b737b2436dfa93cfd3a0c) in order to deal with
comptime_args, optimized_order, and struct and union fields. After
structs and unions get moved into InternPool properly, tmp_hack_arena
can be deleted again.
2023-06-01 23:46:04 -07:00
|
|
|
|
2023-05-08 16:52:59 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-05-10 12:16:24 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-08-24 20:43:43 -07:00
|
|
|
|
2023-05-15 20:09:54 -07:00
|
|
|
|
2023-08-24 20:43:43 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-05-10 12:16:24 -07:00
|
|
|
|
2023-05-10 17:21:22 -07:00
|
|
|
|
2023-08-21 14:27:34 -07:00
|
|
|
|
|
|
|
|
|
2023-05-15 20:09:54 -07:00
|
|
|
|
2023-08-21 14:27:34 -07:00
|
|
|
|
2023-08-24 20:43:43 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-05-10 17:21:22 -07:00
|
|
|
|
stage2: move enum types into the InternPool
Unlike unions and structs, enums are actually *encoded* into the
InternPool directly, rather than using the SegmentedList trick. This
results in them being quite compact, and greatly improved the ergonomics
of using enum types throughout the compiler.
It did however require introducing a new concept to the InternPool which
is an "incomplete" item - something that is added to gain a permanent
Index, but which is then mutated in place. This was necessary because
enum tag values and tag types may reference the namespaces created by
the enum itself, which required constructing the namespace, decl, and
calling analyzeDecl on the decl, which required the decl value, which
required the enum type, which required an InternPool index to be
assigned and for it to be meaningful.
The API for updating enums in place turned out to be quite slick and
efficient - the methods directly populate pre-allocated arrays and
return the information necessary to output the same compilation errors
as before.
2023-05-12 00:07:32 -07:00
|
|
|
|
2023-05-15 20:09:54 -07:00
|
|
|
|
|
|
|
|
|
2023-05-25 05:47:25 -04:00
|
|
|
|
2023-05-15 20:09:54 -07:00
|
|
|
|
|
|
|
|
|
rework generic function calls
Abridged summary:
* Move `Module.Fn` into `InternPool`.
* Delete a lot of confusing and problematic `Sema` logic related to
generic function calls.
This commit removes `Module.Fn` and replaces it with two new
`InternPool.Tag` values:
* `func_decl` - corresponding to a function declared in the source
code. This one contains line/column numbers, zir_body_inst, etc.
* `func_instance` - one for each monomorphization of a generic
function. Contains a reference to the `func_decl` from whence the
instantiation came, along with the `comptime` parameter values (or
types in the case of `anytype`)
Since `InternPool` provides deduplication on these values, these fields
are now deleted from `Module`:
* `monomorphed_func_keys`
* `monomorphed_funcs`
* `align_stack_fns`
Instead of these, Sema logic for generic function instantiation now
unconditionally evaluates the function prototype expression for every
generic callsite. This is technically required in order for type
coercions to work. The previous code had some dubious, probably wrong
hacks to make things work, such as `hashUncoerced`. I'm not 100% sure
how we were able to eliminate that function and still pass all the
behavior tests, but I'm pretty sure things were still broken without
doing type coercion for every generic function call argument.
After the function prototype is evaluated, it produces a deduplicated
`func_instance` `InternPool.Index` which can then be used for the
generic function call.
Some other nice things made by this simplification are the removal of
`comptime_args_fn_inst` and `preallocated_new_func` from `Sema`, and the
messy logic associated with them.
I have not yet been able to measure the perf of this against master
branch. On one hand, it reduces memory usage and pointer chasing of the
most heavily used `InternPool` Tag - function bodies - but on the other
hand, it does evaluate function prototype expressions more than before.
We will soon find out.
2023-07-03 22:09:30 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-07-08 23:39:37 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
rework generic function calls
Abridged summary:
* Move `Module.Fn` into `InternPool`.
* Delete a lot of confusing and problematic `Sema` logic related to
generic function calls.
This commit removes `Module.Fn` and replaces it with two new
`InternPool.Tag` values:
* `func_decl` - corresponding to a function declared in the source
code. This one contains line/column numbers, zir_body_inst, etc.
* `func_instance` - one for each monomorphization of a generic
function. Contains a reference to the `func_decl` from whence the
instantiation came, along with the `comptime` parameter values (or
types in the case of `anytype`)
Since `InternPool` provides deduplication on these values, these fields
are now deleted from `Module`:
* `monomorphed_func_keys`
* `monomorphed_funcs`
* `align_stack_fns`
Instead of these, Sema logic for generic function instantiation now
unconditionally evaluates the function prototype expression for every
generic callsite. This is technically required in order for type
coercions to work. The previous code had some dubious, probably wrong
hacks to make things work, such as `hashUncoerced`. I'm not 100% sure
how we were able to eliminate that function and still pass all the
behavior tests, but I'm pretty sure things were still broken without
doing type coercion for every generic function call argument.
After the function prototype is evaluated, it produces a deduplicated
`func_instance` `InternPool.Index` which can then be used for the
generic function call.
Some other nice things made by this simplification are the removal of
`comptime_args_fn_inst` and `preallocated_new_func` from `Sema`, and the
messy logic associated with them.
I have not yet been able to measure the perf of this against master
branch. On one hand, it reduces memory usage and pointer chasing of the
most heavily used `InternPool` Tag - function bodies - but on the other
hand, it does evaluate function prototype expressions more than before.
We will soon find out.
2023-07-03 22:09:30 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
stage2: move enum types into the InternPool
Unlike unions and structs, enums are actually *encoded* into the
InternPool directly, rather than using the SegmentedList trick. This
results in them being quite compact, and greatly improved the ergonomics
of using enum types throughout the compiler.
It did however require introducing a new concept to the InternPool which
is an "incomplete" item - something that is added to gain a permanent
Index, but which is then mutated in place. This was necessary because
enum tag values and tag types may reference the namespaces created by
the enum itself, which required constructing the namespace, decl, and
calling analyzeDecl on the decl, which required the decl value, which
required the enum type, which required an InternPool index to be
assigned and for it to be meaningful.
The API for updating enums in place turned out to be quite slick and
efficient - the methods directly populate pre-allocated arrays and
return the information necessary to output the same compilation errors
as before.
2023-05-12 00:07:32 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-05-12 16:22:37 -07:00
|
|
|
|
|
|
|
|
|
2023-05-25 05:47:25 -04:00
|
|
|
|
2023-05-12 16:22:37 -07:00
|
|
|
|
rework generic function calls
Abridged summary:
* Move `Module.Fn` into `InternPool`.
* Delete a lot of confusing and problematic `Sema` logic related to
generic function calls.
This commit removes `Module.Fn` and replaces it with two new
`InternPool.Tag` values:
* `func_decl` - corresponding to a function declared in the source
code. This one contains line/column numbers, zir_body_inst, etc.
* `func_instance` - one for each monomorphization of a generic
function. Contains a reference to the `func_decl` from whence the
instantiation came, along with the `comptime` parameter values (or
types in the case of `anytype`)
Since `InternPool` provides deduplication on these values, these fields
are now deleted from `Module`:
* `monomorphed_func_keys`
* `monomorphed_funcs`
* `align_stack_fns`
Instead of these, Sema logic for generic function instantiation now
unconditionally evaluates the function prototype expression for every
generic callsite. This is technically required in order for type
coercions to work. The previous code had some dubious, probably wrong
hacks to make things work, such as `hashUncoerced`. I'm not 100% sure
how we were able to eliminate that function and still pass all the
behavior tests, but I'm pretty sure things were still broken without
doing type coercion for every generic function call argument.
After the function prototype is evaluated, it produces a deduplicated
`func_instance` `InternPool.Index` which can then be used for the
generic function call.
Some other nice things made by this simplification are the removal of
`comptime_args_fn_inst` and `preallocated_new_func` from `Sema`, and the
messy logic associated with them.
I have not yet been able to measure the perf of this against master
branch. On one hand, it reduces memory usage and pointer chasing of the
most heavily used `InternPool` Tag - function bodies - but on the other
hand, it does evaluate function prototype expressions more than before.
We will soon find out.
2023-07-03 22:09:30 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-08-21 14:27:34 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-08-24 20:43:43 -07:00
|
|
|
|
2023-08-21 14:27:34 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-08-24 20:43:43 -07:00
|
|
|
|
|
|
|
|
|
2023-08-21 14:27:34 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-08-24 20:43:43 -07:00
|
|
|
|
2023-08-21 14:27:34 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-08-24 20:43:43 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-08-21 14:27:34 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-08-24 20:43:43 -07:00
|
|
|
|
|
|
|
|
|
2023-08-21 14:27:34 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-08-24 20:43:43 -07:00
|
|
|
|
2023-08-21 14:27:34 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-08-24 20:43:43 -07:00
|
|
|
|
2023-08-21 14:27:34 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-08-24 20:43:43 -07:00
|
|
|
|
2023-08-21 14:27:34 -07:00
|
|
|
|
|
|
|
|
|
2023-08-24 20:43:43 -07:00
|
|
|
|
2023-08-21 14:27:34 -07:00
|
|
|
|
|
|
|
|
|
2023-08-24 20:43:43 -07:00
|
|
|
|
2023-08-21 14:27:34 -07:00
|
|
|
|
|
|
|
|
|
2023-08-24 20:43:43 -07:00
|
|
|
|
|
|
|
|
|
2023-08-21 14:27:34 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-08-24 20:43:43 -07:00
|
|
|
|
2023-08-21 14:27:34 -07:00
|
|
|
|
|
|
|
|
|
2023-08-24 20:43:43 -07:00
|
|
|
|
|
|
|
|
|
2023-08-21 14:27:34 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-08-24 20:43:43 -07:00
|
|
|
|
2023-08-21 14:27:34 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-08-24 20:43:43 -07:00
|
|
|
|
2023-08-21 14:27:34 -07:00
|
|
|
|
|
|
|
|
|
2023-08-24 20:43:43 -07:00
|
|
|
|
2023-08-21 14:27:34 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-08-24 20:43:43 -07:00
|
|
|
|
2023-08-21 14:27:34 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-08-24 20:43:43 -07:00
|
|
|
|
2023-08-21 14:27:34 -07:00
|
|
|
|
2023-08-24 20:43:43 -07:00
|
|
|
|
|
|
|
|
|
2023-08-21 14:27:34 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
sema: support reinterpreting extern/packed unions at comptime via field access
My previous change for reading / writing to unions at comptime did not handle
union field read/writes correctly in all cases. Previously, if a field was
written to a union, it would overwrite the entire value. This is problematic
when a field of a larger size is subsequently read, because the value would not
be long enough, causing a panic.
Additionally, the writing behaviour itself was incorrect. Writing to a field of
a packed or extern union should only overwrite the bits corresponding to that
field, allowing for memory reintepretation via field writes / reads.
I addressed these problems as follows:
Add the concept of a "backing type" for extern / packed unions
(`Type.unionBackingType`). For extern unions, this is a `u8` array, for packed
unions it's an integer matching the `bitSize` of the union. Whenever union
memory is read at comptime, it's read as this type.
When union memory is written at comptime, the tag may still be known. If so, the
memory is written using the tagged type. If the tag is unknown (because this
union had previously been read from memory), it's simply written back out as the
backing type.
I added `write_packed` to the `reinterpret` field of
`ComptimePtrMutationKit`. This causes writes of the operand to be packed - which
is necessary when writing to a field of a packed union. Without this, writing a
value to a `u1` field would overwrite the entire byte it occupied.
The final case to address was reading a different (potentially larger) field
from a union when it was written with a known tag. To handle this, a new kind of
bitcast was introduced (`bitCastUnionFieldVal`) which supports reading a larger
field by using a backing buffer that has the unwritten bits set to
undefined. The reason to support this (vs always just writing the union as it's
backing type), is that no reads to larger fields ever occur at comptime, it
would be strictly worse to have spent time writing the full backing type.
2023-10-01 00:07:21 -04:00
|
|
|
|
2023-08-21 14:27:34 -07:00
|
|
|
|
|
|
|
|
|
2023-09-23 13:03:03 -04:00
|
|
|
|
2023-08-21 14:27:34 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-08-24 20:43:43 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-09-24 14:37:36 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|