Repository navigation
Conversation
release: promote dev into main for v2.32.1
# Conflicts: # package.json
[WRONG BRANCH] merge dev into main for the v2.33.0 release
Promotes the dev integration line onto main. The resulting tree is byte-identical to origin/dev, including package.json at 2.34.0. The package.json conflict is resolved to dev's side, NOT to main's stale 2.33.0. Earlier promotions (lidge-jun#2553, lidge-jun#2507) kept the target's version so the release bump would land on its own "release: vX.Y.Z" commit. That is no longer legal: this very delta adds tests/release-version-line.test.ts, which fails when the in-tree version sits behind the highest release tag. With v2.34.0-preview.20260827 now published, 2.33.0 orders behind it, so a promotion carrying the stale line turns CI red on every shard that runs the suite. The consequence for the release step is that scripts/release.ts skips the bump (release.ts:568, currentVersion === version), so v2.34.0 gets tagged on this merge commit rather than on a separate release commit. The workflow creates the tag itself after publishing and validates expected-sha against the checked-out commit, so the tag still names exactly the audited tree.
[WRONG BRANCH] promote dev onto main for v2.34.0
|
✅ Deterministic PR hygiene checks passed. |
|
Important
This repository does not receive automatic reviews because it has fewer than 10 stars. ⚙️ Run configurationConfiguration used: Path: .coderabbit.yaml Review profile: ASSERTIVE Plan: Team Run ID: Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out. Comment |
github-actions
Bot
changed the title
✅ READY
Hygiene✅ Deterministic PR hygiene checks passed. |
![chatgpt-codex-connector[bot]](https://avatars.githubusercontent.com/in/1144995?s=60&v=4){kind=small}
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: d16b150861
ℹ️ About Codex in GitHub
Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you
- Open a pull request for review
- Mark a draft as ready
- Comment "@codex review".
If Codex has suggestions, it will comment; otherwise it will react with 👍.
Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".
| const replayStart = externalModel | ||
| ? Math.max(0, replayEnd - CURSOR_EXTERNAL_REPLAY_MESSAGE_LIMIT) | ||
| : 0; |
There was a problem hiding this comment.
When more than 4,096 messages follow the most recent historical user message, this cutoff starts on an assistant/tool-result entry. The later complete-turn selector requires a user root, so with the history constructed by the new test (old turn, 4,101 assistant messages, then the active user), it rejects the orphaned suffix and sends only the system root; the external model receives the active “continue” action with none of the conversation being continued. Anchor the bounded suffix at a user/developer boundary, or explicitly retain the preceding user message, so limiting construction cannot erase all replay context.
Useful? React with 👍 / 👎.
luvs01
changed the title
|
Closing evidence (retroactive): this PR was closed without a rationale comment, and the fix was never landed — \�ntries.indexOf(previous.entry)\ is still live on current upstream dev. The change has been re-ported onto current dev as #608 (tests: 9 + 335 pass) and will be submitted upstream when PR creation is permitted again. |
2 participants
Motivation
entries.indexOf(...)for every duplicate, creating attacker-controlled O(k·r) CPU behavior that could stall the server event loop.rawMessagesbefore pruning, so large attacker-controlled histories could trigger the quadratic path before any size limits applied.Description
lastReplayEntryis replaced bylastReplayEntryIndexand duplicate replacements update the known index directly to make each dedupe operation O(1).CURSOR_EXTERNAL_REPLAY_MESSAGE_LIMIT = 4096and limit the synchronous replay construction to the newest bounded suffix for external wire models so replay processing is bounded before later wire-size pruning.bounds replay construction before processing an oversized historytotests/cursor-repetition-breaker.test.tsthat verifies oversized histories outside the bounded replay window do not trigger the severe repetition note.Testing
bun test tests/cursor-repetition-breaker.test.tsand all 6 tests passed.bun run typecheck(TS check) which completed without errors.bun run test; the repository-wide suite ran many tests and reported unrelated failures/timeouts in other subsystems, so only the focused regression and typecheck were relied on to validate this scoped fix.Codex Task