perf(oxlint): align walker thread count with rayon pool#22494
Merged
Conversation
camc314
approved these changes
May 17, 2026
Contributor
Merge activity
|
## Summary - The file walker (`ignore::WalkParallel`) was using its own default thread count (`available_parallelism()`), independent of rayon's pool size. When users pass `--threads=N` to lower rayon's count, the walker still silently spun up a full background pool, adding kernel pressure during the walk phase. - This brings oxlint in line with oxfmt, which already does the same: `apps/oxfmt/src/cli/walk.rs:416`. ## Measurement On a 72k-file repo (rolldown) on M3 Max, 8 runs per config: | Config | Wall | User | Sys | |---|---|---|---| | `--threads=14` (default) before | 3.31 s | 2.50 s | 32.0 s | | `--threads=14` after | 3.21 s | 2.50 s | 29.4 s | | `--threads=6` before | 1.36 s | 1.88 s | 5.88 s | | `--threads=6` after | **1.28 s** | 1.84 s | **3.91 s** | The win is largest when `--threads` is set below `available_parallelism()`: 33% sys-time reduction at `--threads=6`. At default threads it's a small consistency improvement (no behavior change since both pools were already at the same count). 🤖 Generated with [Claude Code](https://claude.com/claude-code)
5e89ea8 to
4c9ca72
Compare
camc314
pushed a commit
that referenced
this pull request
May 18, 2026
# Oxlint ### 🚀 Features - 1ae291e linter/no-underscore-dangle: Add `allowInUsingDeclarations` option (#22483) (吴杨帆) - 0440b0f linter/eslint: Implement `id-match` rule (#22379) (Vladislav Sayapin) - 65bf119 linter: Implement react no-object-type-as-default-prop (#22481) (uhyo) - 2a6ddce linter/eslint: Implement `no-implied-eval` rule (#22391) (Vladislav Sayapin) - d3a3c1d linter: Auto detect agents from CLI and transition to the agent output format (#22068) (Jovi De Croock) - 625758a linter/vitest: Implement padding-around-after-all-blocks rule (#21788) (kapobajza) - 37680b0 linter: Implement react no-unstable-nested-components (#22248) (Jovi De Croock) - d8d9c74 linter: Implement import/newline-after-import rule (#19142) (Ryuya Yanagi) ### 🐛 Bug Fixes - 3f59e03 linter: Only call rayon/miette/tracing inits once (#21899) (Matiss Janis Aboltins) - 602dfd6 linter/promise/no-return-wrap: Detect Promise calls in all branches (#22474) (zennnnnnn11) - e182aee linter: Allow dialogs and popovers for no_autofocus (#22289) (mehm8128) - 7ffb710 linter/jest/vitest: Jest/no-standalone-expect ignores additionalTestBlockFunctions option for jest/vitest hooks (#22477) (kapobajza) - c6f2d3f linter: Add more expression support for iframe-has-title (#22460) (mehm8128) - 5747ff1 linter: Avoid enabling jest with vitest plugin (#22499) (camc314) - 863984f linter/no-find-dom-node: Run on all files (#22479) (bab) ### ⚡ Performance - 2afef79 linter: Optimize `no-loop-func` (#22491) (camchenry) - 4c9ca72 oxlint: Align walker thread count with rayon pool (#22494) (Boshen) ### 📚 Documentation - f7967c7 linter/id-match: Clarify `onlyDeclarations` config docs (#22523) (camc314) - 1e0c97f linter: Fix closing code block in documentation for `padding-around-after-all-blocks` rule. (#22513) (connorshea) - a9049fd linter: Exclude directly provide autoFocus to dialog pattern (#22510) (mehm8128) # Oxfmt ### 🐛 Bug Fixes - 8ee946f formatter/sort_imports: Use label to classify lines (#22512) (leaysgur) - 8c1da44 formatter: Normalize destructuring keys in DCR (#22478) (camc314)
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary
ignore::WalkParallel) was using its own default thread count (available_parallelism()), independent of rayon's pool size. When users pass--threads=Nto lower rayon's count, the walker still silently spun up a full background pool, adding kernel pressure during the walk phase.apps/oxfmt/src/cli/walk.rs:416.Measurement
On a 72k-file repo (rolldown) on M3 Max, 8 runs per config:
--threads=14(default) before--threads=14after--threads=6before--threads=6afterThe win is largest when
--threadsis set belowavailable_parallelism(): 33% sys-time reduction at--threads=6. At default threads it's a small consistency improvement (no behavior change since both pools were already at the same count).🤖 Generated with Claude Code