[FR]: Browser Windows Should Have Roles and Weights: A Design Idea #2549

Closed
opened 2026-09-21 06:02:47 +00:00 by smart-z · 1 comment
smart-z commented 2026-09-21 06:02:47 +00:00 (Migrated from github.com)

Description

Browser windows should be like different workspaces with their own roles and priorities, keeping the main one focused while temporary content goes where it belongs, all shaped by what you’re doing.

Who's implementing?

  • I'm willing to implement this feature myself

The problem

Browser Windows Should Have Roles and Weights: A Design Idea

This post is to share a feeling I’ve had after using browsers for a long time. It’s not a feature request, more of a design direction. If it resonates, I’d be happy to discuss.

The problem I run into

I usually keep a main browser window on the left, basically fixed, for my main workflow. On the right, I keep a secondary window for temporary searches, external links, and quick lookups. Some high-frequency apps become PWAs on the desktop. Ideally, I’d also have a window tucked at the edge of the screen that slides out when I move my mouse there.

But browsers treat all of these as the same kind of window. A new link steals focus. A quick search interrupts the main workflow. External links enter the main tab system and pollute the current context. Then I have to Ctrl+Tab back.

Arc’s Little Arc made me realize something: it’s useful not because it’s “small,” but because it’s low-weight — temporary, doesn’t steal focus, doesn’t enter the main tab system, previewable, promotable, disposable.

Core idea: separate form and weight

A window has two dimensions:

  • Form: what it looks like. Main window, overlay, edge drawer, PWA, background window.
  • Weight: how it exists. Does it steal focus? Does it show in the taskbar? How long does it live? Does it enter the tab system? Can it be promoted or demoted?

Today browsers mostly distinguish only “window” and “tab,” without defining window roles. So by default, every new window becomes the same thing: steals focus, takes a taskbar slot, enters the tab system, and needs manual management.

If the “low-weight” idea behind Little Arc became a general capability, and different windows could have different weights, this would feel much better.

Window roles

  • Anchor: long-lived, focus-protected, not easily replaced.
  • Scratch: receives temporary links and searches, doesn’t steal focus by default, cleaned up when idle.
  • Peek: the Little Arc kind — Esc to close, no taskbar presence.
  • Shelf: tucked at the screen edge, slides out, for notes, calendar, references.
  • PWA / App Window: standalone icon, long-lived, shares selected identity.
  • Ghost: opens in the background, doesn’t steal focus, can be checked later.

Key capabilities: which role new links go to by default; whether they can open in the background without stealing focus; whether they can be promoted to Anchor or pinned as a PWA with one action; whether they can be demoted, archived, or moved across windows; whether identity is shared cookies or isolated.

Contexts

Right now Profile, Workspace, Window, Little Arc, PWA, and Air Traffic Control are fragmented: Profile handles identity, Workspace handles tab grouping, Window handles display, Little Arc handles low-weight temporary content, PWA handles desktop-ification, and Air Traffic Control handles part of routing.

But what users actually face is a task context: work, reading, temporary. In that context, identity, windows, routing, focus, lifecycle, and layout are all connected.

For example, a “Work context”: Anchor on the left with company dashboards, email, and docs; Scratch on the right for external links, quick searches, and ChatGPT; Shelf for calendar and todos; Slack and Notion as PWAs in the Dock; company domains go to Anchor, external docs go to Scratch, searches go to Peek; Scratch doesn’t steal focus from Anchor by default.

A “Reading context” looks different: Anchor for the article, Shelf for notes, Peek for dictionary lookups, PWAs for read-later and Readwise.

This is closer to real tasks than just managing profiles.

Understanding it as layers

It helps to break it into L0–L5:

  • L0 Identity: who I am, which account, which cookie set. Corresponds to Profile / Container.
  • L1 Context: what I’m doing — work, reading, temporary. This is the top layer.
  • L2 Window role: whether this window is Anchor, Scratch, Peek, Shelf, or PWA.
  • L3 Routing: which window role a link should go to. An upgraded Air Traffic Control.
  • L4 Focus and lifecycle: whether it steals focus, appears in the taskbar, how long it lives, when it closes.
  • L5 Layout: which screen, which position, how large.

Today’s products cover only some of these layers, and they don’t connect. Profile covers L0. Workspace covers part of L1. Window covers L5. Little Arc covers part of L2 and L4. Ideally, one “context” would include all of them.

Routing can be more fine-grained

Based on source, URL, and current context, decide: which window role, whether to steal focus, which identity, how long it lives, and which screen it goes to.

For example:

External app link -> Scratch, no focus steal
Company domain -> Anchor, focus steal
Search / temporary -> Peek
docs.google.com -> Anchor or PWA
youtube.com -> Scratch, muted, no focus steal

Minimum prototype

If starting with a minimum version:

  • Two window roles: Anchor + Scratch.
  • New external links go to Scratch by default, without stealing focus.
  • A shortcut promotes Scratch to Anchor.
  • A Little Arc-style Peek handles search and preview.
  • PWAs are standalone icons and share identity.
  • A simple routing table.

Prove that “window roles + focus protection + low-weight temporary windows” are valuable first, then build up.

Questions to discuss

  • Should window roles be explicit, or inferred from routing rules?
  • Should Scratch appear in the taskbar? Or just show a badge on Anchor?
  • Should identity isolation follow context or window role?
  • Can Shelf and PWA share one window management API?
  • How do we avoid routing rules becoming yet another config to maintain?

One-line summary

I don’t want more windows. I want a window system with roles, weights, and routing rules. Generalize Little Arc’s “low-weight” idea, then use contexts to unify identity, routing, focus, and lifecycle.

If this direction seems useful, I’d love to hear your thoughts.

Disclosure: This article was polished with AI assistance.

Possible solutions

This is an idea-level proposal, not a concrete spec. The core direction is to give windows roles and weights, keep the main window focus-protected, and route new links by context to Scratch / Peek / Shelf / PWA / Ghost. A minimal starting point could be: Anchor + Scratch, background open without focus steal, a shortcut to promote Scratch to Anchor, and a simple routing table.

Additional context

No response

### Description Browser windows should be like different workspaces with their own roles and priorities, keeping the main one focused while temporary content goes where it belongs, all shaped by what you’re doing. ### Who's implementing? - [ ] I'm willing to implement this feature myself ### The problem # Browser Windows Should Have Roles and Weights: A Design Idea This post is to share a feeling I’ve had after using browsers for a long time. It’s not a feature request, more of a design direction. If it resonates, I’d be happy to discuss. ## The problem I run into I usually keep a main browser window on the left, basically fixed, for my main workflow. On the right, I keep a secondary window for temporary searches, external links, and quick lookups. Some high-frequency apps become PWAs on the desktop. Ideally, I’d also have a window tucked at the edge of the screen that slides out when I move my mouse there. But browsers treat all of these as the same kind of window. A new link steals focus. A quick search interrupts the main workflow. External links enter the main tab system and pollute the current context. Then I have to `Ctrl+Tab` back. Arc’s Little Arc made me realize something: it’s useful not because it’s “small,” but because it’s **low-weight** — temporary, doesn’t steal focus, doesn’t enter the main tab system, previewable, promotable, disposable. ## Core idea: separate form and weight A window has two dimensions: - **Form**: what it looks like. Main window, overlay, edge drawer, PWA, background window. - **Weight**: how it exists. Does it steal focus? Does it show in the taskbar? How long does it live? Does it enter the tab system? Can it be promoted or demoted? Today browsers mostly distinguish only “window” and “tab,” without defining window roles. So by default, every new window becomes the same thing: steals focus, takes a taskbar slot, enters the tab system, and needs manual management. If the “low-weight” idea behind Little Arc became a general capability, and different windows could have different weights, this would feel much better. ## Window roles - **Anchor**: long-lived, focus-protected, not easily replaced. - **Scratch**: receives temporary links and searches, doesn’t steal focus by default, cleaned up when idle. - **Peek**: the Little Arc kind — Esc to close, no taskbar presence. - **Shelf**: tucked at the screen edge, slides out, for notes, calendar, references. - **PWA / App Window**: standalone icon, long-lived, shares selected identity. - **Ghost**: opens in the background, doesn’t steal focus, can be checked later. Key capabilities: which role new links go to by default; whether they can open in the background without stealing focus; whether they can be promoted to Anchor or pinned as a PWA with one action; whether they can be demoted, archived, or moved across windows; whether identity is shared cookies or isolated. ## Contexts Right now Profile, Workspace, Window, Little Arc, PWA, and Air Traffic Control are fragmented: Profile handles identity, Workspace handles tab grouping, Window handles display, Little Arc handles low-weight temporary content, PWA handles desktop-ification, and Air Traffic Control handles part of routing. But what users actually face is a task context: work, reading, temporary. In that context, identity, windows, routing, focus, lifecycle, and layout are all connected. For example, a “Work context”: Anchor on the left with company dashboards, email, and docs; Scratch on the right for external links, quick searches, and ChatGPT; Shelf for calendar and todos; Slack and Notion as PWAs in the Dock; company domains go to Anchor, external docs go to Scratch, searches go to Peek; Scratch doesn’t steal focus from Anchor by default. A “Reading context” looks different: Anchor for the article, Shelf for notes, Peek for dictionary lookups, PWAs for read-later and Readwise. This is closer to real tasks than just managing profiles. ## Understanding it as layers It helps to break it into L0–L5: - **L0 Identity**: who I am, which account, which cookie set. Corresponds to Profile / Container. - **L1 Context**: what I’m doing — work, reading, temporary. This is the top layer. - **L2 Window role**: whether this window is Anchor, Scratch, Peek, Shelf, or PWA. - **L3 Routing**: which window role a link should go to. An upgraded Air Traffic Control. - **L4 Focus and lifecycle**: whether it steals focus, appears in the taskbar, how long it lives, when it closes. - **L5 Layout**: which screen, which position, how large. Today’s products cover only some of these layers, and they don’t connect. Profile covers L0. Workspace covers part of L1. Window covers L5. Little Arc covers part of L2 and L4. Ideally, one “context” would include all of them. ## Routing can be more fine-grained Based on source, URL, and current context, decide: which window role, whether to steal focus, which identity, how long it lives, and which screen it goes to. For example: ``` External app link -> Scratch, no focus steal Company domain -> Anchor, focus steal Search / temporary -> Peek docs.google.com -> Anchor or PWA youtube.com -> Scratch, muted, no focus steal ``` ## Minimum prototype If starting with a minimum version: - Two window roles: Anchor + Scratch. - New external links go to Scratch by default, without stealing focus. - A shortcut promotes Scratch to Anchor. - A Little Arc-style Peek handles search and preview. - PWAs are standalone icons and share identity. - A simple routing table. Prove that “window roles + focus protection + low-weight temporary windows” are valuable first, then build up. ## Questions to discuss - Should window roles be explicit, or inferred from routing rules? - Should Scratch appear in the taskbar? Or just show a badge on Anchor? - Should identity isolation follow context or window role? - Can Shelf and PWA share one window management API? - How do we avoid routing rules becoming yet another config to maintain? ## One-line summary I don’t want more windows. I want a window system with roles, weights, and routing rules. Generalize Little Arc’s “low-weight” idea, then use contexts to unify identity, routing, focus, and lifecycle. If this direction seems useful, I’d love to hear your thoughts. Disclosure: This article was polished with AI assistance. ### Possible solutions This is an idea-level proposal, not a concrete spec. The core direction is to give windows roles and weights, keep the main window focus-protected, and route new links by context to Scratch / Peek / Shelf / PWA / Ghost. A minimal starting point could be: Anchor + Scratch, background open without focus steal, a shortcut to promote Scratch to Anchor, and a simple routing table. ### Additional context _No response_
helium-gardener[bot] commented 2026-09-21 06:08:19 +00:00 (Migrated from github.com)

This issue appears to combine separate feature requests. Could you keep this issue focused on one request and open another issue for the other? If we have misunderstood the scope, please let us know.

This issue appears to combine separate feature requests. Could you keep this issue focused on one request and open another issue for the other? If we have misunderstood the scope, please let us know. <!-- helium-scope-split -->
Sign in to join this conversation.
No milestone
No assignees
1 participant
Notifications
Due date
The due date is invalid or out of range. Please use the format "yyyy-mm-dd".

No due date set.

Dependencies

No dependencies set.

Reference
imputnet/helium#2549
No description provided.