A re-run of phase-pr on a green phase writes ci_pending but keeps the old ci_gated_sha #608

New issue

Closed

opened 2026-09-22 06:25:45 +00:00 by claude · 2 comments

claude commented

2026-09-22 06:25:45 +00:00

Member

Found during IW-601 phase 4 (analysis.md re-measure, §3.2).

handOffWait (PhasePr.scala:388-409) and markPending (CiGate.scala:376-390) write ci_pending without clearing ci_gated_sha, so a re-run of phase-pr on an already-green phase leaves ci_gated_sha pointing at a commit that no longer describes the head the new wait is for.

Related: #568 (one builder for the ci_pending card would clear ci_gated_sha in one place).

Found during IW-601 phase 4 (`analysis.md` re-measure, §3.2). `handOffWait` (`PhasePr.scala:388-409`) and `markPending` (`CiGate.scala:376-390`) write `ci_pending` without clearing `ci_gated_sha`, so a re-run of `phase-pr` on an already-green phase leaves `ci_gated_sha` pointing at a commit that no longer describes the head the new wait is for. Related: #568 (one builder for the `ci_pending` card would clear `ci_gated_sha` in one place).