Repository navigation
perf(worker): avoid quadratic fd tracking - #188
Merged
Merged
Conversation
Codex Review SummaryThis comment shows the latest Codex review activity on this pull request.
ℹ️ About Codex in GitHubYour team has set up Codex to review pull requests in this repo. Reviews are triggered when you
Codex reacts with 👀 while any review is running, comments if it has suggestions, and reacts with 👍 once all reviews finish with no findings. |
This was referenced Oct 17, 2026
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
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
#184 tracks Worker-owned file descriptors in a
Vec, with linearcontainsandpositionscans. A synchronous 2,000-fd open/close loop in a Worker was 18.74% slower than before #184; the bootstrap 95% CI for the paired difference excludes zero.Replace the vector with hash-set tracking, preserving insert/remove semantics. Exit-time close order is irrelevant, and Node's
trackUnmanagedFdsalso uses an unordered set.The loop is 19.7% faster than
785be5fefand 1.95% above the pre-#184 baseline. The remaining ~0.18 ms nested-exit cost is the intended work of closing descriptors that pre-#184 leaked.Validation: all 15 ownership tests in
test/js/node/worker_threads/worker-track-unmanaged-fds.test.tspass.bench/worker-fd-tracking.mjsprovides the regression benchmark; formatting and whitespace checks pass.