Repository navigation
Hold pr_review.py merge for a Hosted Review Whose Status Has Not Registered Yet #3806
Description
#3789's fix holds pr_review.py wait and refuses merge while a CodeRabbit or Qodo check on the head is still settling. It sees only a check that has already registered on the head.
#3789 names a second factor that this does not cover. CodeRabbit's status can register several minutes after the other checks finish. A head whose required checks all pass before CodeRabbit posts its pending status reads CLEAN or UNSTABLE with no hosted check in the rollup. So merge can still land before the review starts. That failure is silent.
Closing it needs a signal that a hosted reviewer is going to review this head. Nothing on the pull request states that today. Candidate signals:
- The reviewer reviewed an earlier head of the same pull request, so a review of this head is expected.
- The repository's reviewer configuration says the reviewer runs on this base branch.
- A short grace period after the last check finishes, read against the head's push time.
Each has a false-hold cost on a pull request the reviewer skips, such as an incremental review CodeRabbit declines. Choosing one is a design decision.
Activity
Hold pr_review.py merge on a Running Hosted Review and Route Declined…
ptr727 commentedon Nov 2, 2026
Held back by the backlog-burndown away run (2026-11-02, round 2): the signal choice is filed as decision #3814. This issue is worked once that is answered.
Fix Adjacent Docs and Review Tails in the Open Pull Request Instead o…
ptr727 commentedon Nov 3, 2026
Maintainer answer on #3814 (decision triage via the windows handoff #3853, 2026-11-03): Option 1: a grace period after the later of the head's push and its last check settling, then release if no hosted check has registered. The length is not set yet: the change closing this issue measures registration delay on recent pull requests and proposes a value from that, for the maintainer to approve in its pull request.