Repository navigation
merge: auto merges a cleared head while its author is still at work - #601
Merged
Merged
Conversation
The self-merge waited until the run had no jobs in flight and no unread messages. An author that answered each panel round's advisory findings with another experiment never went quiet, so its cleared PR stayed open for days while main fell behind. Only an unread human comment holds the merge now; the author's brief says a cleared PR merges without waiting for it.
There was a problem hiding this comment.
Round 2 (re-run on the same head) — reviewed head 5dc93cf4 — reviewer summarizer:hermes/gpt-5.6-terra over coverage+credentials+deployment+general+lifecycle+prose.
terra
Advisory findings from outerloop — the code owner decides. Reply to disagree; the outerloop:no-review label opts this PR out.
Verdict: 1 blocking, 0 advisory.
1 finding attached to the lines below.
Merged verdict: one blocking lifecycle finding. Rejected findings: none.
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Need
Under
merge: auto, the kernel merged a blessed head only once the run was parked with no jobs in flight and no unread messages. Every review round produces advisory findings, and each one arrives as a message. An author that answers them with another experiment never goes quiet. On one live target, a PR cleared the gate and the panel with no blocking findings, then stayed open for nine days. Its author kept revising, which also improved the score, but main never received any of it.Change
_merge_blessed_prno longer waits on the author's jobs or on the loop's own messages (verdicts, launch results, CI). It still waits on an unread human comment (human_pending, sourcehuman). Every other check is unchanged: the run must be parked, the head must be the blessed head, the head must contain the base tip, CI must be clean, and the contract must say auto.finish_run). Further work comes back as a new PR.Compatibility: no persisted state changes. Runs in flight pick up the new rule at the next tick, so a cleared PR whose author is still running jobs merges on that tick, and those jobs are cancelled.
Tests: the self-merge parametrizations now cover a busy author (merges) and a human comment (waits). The rebless and lease-race cases use a human comment where they previously used a kernel note.
Built by Claude Opus 5.5; review by the PR review bot.