Repository navigation
Fix targeted review continuation and configurable execution timeouts - #59
Merged
Merged
Conversation
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Summary
Large reviews previously re-sent whole groups during continuation and could spend their execution budget before later groups or synthesis. This change checkpoints assigned files, orders changed dependencies together, gives every group a first pass, and continues only unfinished work without discarding validated findings.
review.timeout_minutesand--timeout MINUTESshare one host-enforced deadline and the unchanged global 100-request allowance. The GitHub-hosted six-hour job ceiling reserves 30 minutes for setup/refresh/cleanup. Host-only serialized token renewal persists encrypted rotation before updating the proxy-only access token. Read-token renewal repeats live OIDC/installation/permission/commit checks. Request expiry and leases follow the chosen timeout; merge intent still expires independently after one hour.Evidence
After: deterministic regressions check positive timeout values, precedence, equivalent-command idempotency, request/lease expiry, concurrency, assigned-file coverage, dependency grouping, pending-only continuation, synthesis reserve and fail-closed output.
npm run check: 267 passing tests, typecheck, lint and build. Clean source install/check, real workerd/D1, synthetic browser onboarding, pinned model catalog, audit (zero vulnerabilities), Actionlint and Wrangler dry-run passed.--timeout 60. Native COMMENT publication is bound to each exact head. The 93-file fixture has 1,668 changed lines, not the roughly 8,700 in the production report; this does not guarantee every large PR completes. A separate >12,000-line planner regression checks bounded lossless grouping.23018028dc656100c1b313a01bf20e7109b8258a; credential-only production OIDC checks passed with inference skipped. Regular release v0.1.3 is deployed with the merge gate restored, browser/hosted-inference flags disabled, and health verified. Test credentials and resources were cleaned up; no KMU AI review was triggered.Merge Danger
Door: two-way
No database migration. Runner and Worker changes require coordinated immutable source/caller pins after successful isolated acceptance; do not deploy the runner alone.
Blast Radius: review-runtime
Longer operator-selected reviews consume additional Actions/ChatGPT allowance. The reviewer remains network-isolated, read-only and credential-free; incomplete/unverified analysis never approves or merges. Token rotation failure remains fail-closed and may require reconnection after an uncertain provider exchange.