perf: cache-key and hot-path improvements atop issue #449 normalization#532
Merged
alexander-akait merged 2 commits intomainfrom Apr 17, 2026
Merged
Conversation
Extends the cache-hit work from PR #481 with additional hot-path wins that don't change public behavior or semantics: - UnsafeCachePlugin.getCacheId: build the cache id by direct string concat with a `\0` separator instead of `JSON.stringify`ing a 6-key object per lookup. Paths, requests, queries and fragments produced by parseIdentifier never contain raw `\0`; context is still JSON-stringified when included, which escapes any `\0` there. - util/path.isRelativeRequest: char-code-based check that replaces the `/^\.\.?(?:\/|$)/` regex used in UnsafeCachePlugin and getInnerRequest. - DescriptionFileUtils.cdUp: single reverse char-code scan instead of two `lastIndexOf` calls (POSIX strings don't need to be fully re-scanned for a backslash that isn't there). - DescriptionFilePlugin: skip the global-regex backslash replace when the relative slice doesn't contain any backslashes. Also adds `benchmark/cases/unsafe-cache-miss-heavy/` which exercises the cache-id path directly (a miss pass followed by a hit pass over 42 unique requests, so `getCacheId` is the dominant cost) and unit tests for `isRelativeRequest`. Local results (Node 22, warm fs cache): unsafe-cache (ON, 3x repeat): 11.5k → 14.5k ops/s (+25%) unsafe-cache-miss-heavy: 2.2k → 2.8k ops/s (+27%) realistic-midsize (warm cache): 4.2k → 4.6k ops/s (+9%) realistic-midsize (cold cache): 42.5 → 46.6 ops/s (+10%) One internal change: UnsafeCachePlugin cache keys are no longer valid JSON, so test/yield.test.js was updated to split on the `\0` separator instead of `JSON.parse`ing a key. The cache-key format has never been part of the public API. https://claude.ai/code/session_014HMqKsoba7AvAVojZWECzz
|
Codecov Report✅ All modified and coverable lines are covered by tests. Additional details and impacted files@@ Coverage Diff @@
## main #532 +/- ##
==========================================
+ Coverage 96.52% 96.58% +0.05%
==========================================
Files 50 50
Lines 2651 2665 +14
Branches 808 812 +4
==========================================
+ Hits 2559 2574 +15
+ Misses 77 76 -1
Partials 15 15
Flags with carried forward coverage won't be shown. Click here to find out more. ☔ View full report in Codecov by Sentry. 🚀 New features to boost your workflow:
|
Merging this PR will not alter performance
Performance Changes
Comparing 
|
|
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.
Extends the cache-hit work from PR #481 with additional hot-path wins that
don't change public behavior or semantics:
with a
\0separator instead ofJSON.stringifying a 6-key object perlookup. Paths, requests, queries and fragments produced by parseIdentifier
never contain raw
\0; context is still JSON-stringified when included,which escapes any
\0there./^\.\.?(?:\/|$)/regex used in UnsafeCachePlugin and getInnerRequest.lastIndexOfcalls (POSIX strings don't need to be fully re-scanned fora backslash that isn't there).
relative slice doesn't contain any backslashes.
Also adds
benchmark/cases/unsafe-cache-miss-heavy/which exercises thecache-id path directly (a miss pass followed by a hit pass over 42 unique
requests, so
getCacheIdis the dominant cost) and unit tests forisRelativeRequest.Local results (Node 22, warm fs cache):
unsafe-cache (ON, 3x repeat): 11.5k → 14.5k ops/s (+25%)
unsafe-cache-miss-heavy: 2.2k → 2.8k ops/s (+27%)
realistic-midsize (warm cache): 4.2k → 4.6k ops/s (+9%)
realistic-midsize (cold cache): 42.5 → 46.6 ops/s (+10%)
One internal change: UnsafeCachePlugin cache keys are no longer valid JSON,
so test/yield.test.js was updated to split on the
\0separator insteadof
JSON.parseing a key. The cache-key format has never been part of thepublic API.
https://claude.ai/code/session_014HMqKsoba7AvAVojZWECzz