Repository navigation
π stale-red base refresh: update PRs returned to their contributor (072) - #18949
hivecommons-hive[bot] merged 1 commit into
Conversation
The stale-red base refresh merges the base into a red PR only when every failing check is green on the base tip, so the red it clears is the base's. It still skipped PRs at stage/072-returned-to-contributor on the assumption that the contributor's hive pushes there, but nothing asks the contributor to merge the base: projectbluefin/actions#765 and hivecommons#766 stayed red at 072 for ten hours after main fixed their checks in hivecommons#768 (both green on the base tip all night), until a person pressed "Update branch". The 072 skip is lifted for the stale-red refresh only; the missing-check refresh keeps it. The update stays pinned to the evaluated head, so a push from the contributor wins. Signed-off-by: Danathar <6772335+Danathar@users.noreply.github.com> Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01BQYnPGFJx81c8JXR3o1sg1
|
[APPROVALNOTIFIER] This PR is NOT APPROVED This pull-request has been approved by: The full list of commands accepted by this bot can be found here. DetailsNeeds approval from an approver in each of these files:Approvers can indicate their approval by writing |
|
@Danathar this PR is back in your court ( Review asks:
Reply here or push a new head to this branch (same sign-off); CI and the review swarm re-run automatically and the PR returns to the normal path. Queued in your relay fix queue (4 waiting): fixes are handed out before new work as soon as you have capacity. |
Open PR overlap checkBase branch: Conflicts are checked with every PR merged onto the current tip of
|
Summary
The stale-red base refresh (from #17544) skipped PRs at
stage/072-returned-to-contributor. It assumed the contributor's hive would bring those branches up to date. Nothing asks it to, so a 072 PR whose red the base had since fixed stayed red until a person pressed "Update branch". This PR lets the stale-red refresh update 072 PRs. Every other skip is unchanged, and the missing-check refresh keeps its 072 skip.What happened
Audit GitHub Actions(andpyteston fix: remove NO_GITHUB mode, use ADVISORY as lowest tierΒ #765), because main had a stale zizmor baseline. Both moved to 072 at about 4:20 PM EST on 11-17.75348afb3stayed the tip all night, and it ranAudit GitHub Actionsandpytestgreen. So the refresh's own test, "every failing check is green on the base tip", was met for both PRs.stale-red base refresh skippedline for either PR between 8:00 PM and 6:05 AM. That fits:staleRedRefreshSkipreturnedreturned-to-contributor, which is filtered before anything is logged.The change
staleRedRefreshSkipno longer skips 072. The pass still updates a PR only when the base tip clears every failing check, so the red it clears is the base's, not the contributor's.baseRefreshBranchOwnerSkipis unchanged, so the missing-check refresh still skips 072. A 072 PR there is usually waiting on review changes, not on the base.What I checked and did not change
The batch prompt also suggested the refresh misses checks that run only on PRs, because they never appear on the base tip. That's true in the code:
StaleRedBaseVerdictForreturns base-unknown for a check with no completed run on the tip, as #17544 chose. But it isn't what held these two PRs. Both failing checks ran green on the actions main tip, and the prompt's claim that the tip runs only four checks doesn't match the tip's check runs. So this PR doesn't add open-PR evidence to the stale-red verdict. That can be a separate change if a case shows up.Tests
TestStaleRedRefreshSkip: a 072 PR is now a candidate.returned-to-contributor.Locally I ran only
gofmt -l(AGENTS.md bans localgo test/go build/go vet) andcheck-docs-citations.py, which verified 2163 references. CI is the test run.Appendix: code refs
src/cmd/hive/stale_red_base_refresh.go:staleRedRefreshSkip, and the comments onrefreshStaleRedBehindPRsandbaseRefreshBranchOwnerSkipsrc/cmd/hive/stale_red_base_refresh_test.gosrc/docs/operator-reference/sweeps-and-operations.md: the stale-red section and the write-surface table rowπ€ Generated with Claude Code
https://claude.ai/code/session_01BQYnPGFJx81c8JXR3o1sg1