mirror of
https://github.com/TanStack/router.git
synced 2026-11-16 19:38:07 +00:00
|
History
|
||
|---|---|---|
| .. | ||
| scripts | ||
| src/fixtures | ||
| tests | ||
| eslint.config.js | ||
| eslint.perf-async.config.js | ||
| eslint.perf-baseline.config.js | ||
| eslint.perf-both.config.js | ||
| eslint.perf-server.config.js | ||
| package.json | ||
| README.md | ||
| tsconfig.json | ||
| vitest.config.ts | ||
ESLint Plugin Start (e2e)
Runs end-to-end linting tests against @tanstack/eslint-plugin-start.
Prereqs
pnpm installat repo roothyperfineinstalled (used for perf benchmarking)
Tests
- Unit-style e2e via Vitest:
pnpm -C e2e/eslint-plugin-start test:e2e - Lint all fixtures:
pnpm -C e2e/eslint-plugin-start test:lint
Perf benchmark (hyperfine)
Generates synthetic fixtures (gitignored) and measures lint time.
Modes
small: no noise files, just a route + component chain. Measures the cost of the "real" rule work.huge: thousands of irrelevant files + a route + component chain. Measures gating overhead across a large codebase.
Bench commands
pnpm -C e2e/eslint-plugin-start perf:bench:smallpnpm -C e2e/eslint-plugin-start perf:bench:huge
Optional env vars (huge mode):
PERF_NOISE_FILES=5000(default)PERF_CHAIN_DEPTH=50(default)
Notes:
perf:lintonly targetssrc/perf/generated/**so results aren’t dominated by other fixtures.- Use a fresh terminal and close editors/watchers for stable numbers.
- If you want to compare versions, run the same command after each change and keep env vars constant.
- Optional profiling (writes a
*.cpuprofilefile):node --cpu-prof --cpu-prof-name=cpu.cpuprofile ./node_modules/eslint/bin/eslint.js "src/perf/generated/**/*.{ts,tsx}"