build: stop running PR title check on synchronize - #203
Merged
tokoko merged 1 commit intoJul 14, 2026
Conversation
The PR title check only validates the PR title and description, neither of which changes on a push (synchronize). Including synchronize caused a race with Dependabot rebases, which force-push and edit the PR body at the same time, firing synchronize and edited together. Both events land in the same concurrency group, so one run is cancelled, leaving a cancelled check-run on the head commit that fails the status rollup even though a successful run exists. Trigger only on opened, edited, and reopened -- the events that actually change the title or description.
nielspardon
requested review from
andrew-coleman,
benbellick,
bestbeforetoday,
tokoko and
vbarua
July 14, 2026 08:10
tokoko
approved these changes
Jul 14, 2026
tokoko
pushed a commit
that referenced
this pull request
Jul 20, 2026
## Problem The **PR title / description conforms to semantic-release** check — a required status check on `main` — stops being reported once a PR receives a new commit, leaving the PR permanently `BLOCKED` (e.g. #224). The check runs and passes when the PR is opened, but pushing a follow-up commit never re-reports it. ## Root cause GitHub keys required status checks to the **head commit** of the PR. #203 dropped `synchronize` from the workflow triggers (to fix the double check-run race in #198), so a push no longer runs the check. The passing check-run stays attached to the *previous* head commit, and the new head commit has no `PR title / description conforms to semantic-release` run at all — so the status rollup can never be satisfied. Dropping `synchronize` is fundamentally incompatible with this being a required check: any contributor who pushes after opening a PR strands the required check. ## Fix - Re-add `synchronize` so the check reports on every head commit. - Set `cancel-in-progress: false` so the #198 race doesn't return. That race left a *cancelled* check-run because `synchronize` and `edited` (fired together by a Dependabot rebase on the same commit) landed in the same concurrency group and one cancelled the other. With cancellation disabled the second event queues behind the first and both finish as successes — two harmless `SUCCESS` check-runs instead of one `SUCCESS` + one `CANCELLED`. The concurrency group is kept so runs for a PR still serialize and the comment job can't race with itself. 🤖 Generated with AI
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
2 participants
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.
Problem
The PR Title Check registers as failing on some PRs even though the title conforms — see #198. The check ends up with two check-runs of the same name on the head commit, one
SUCCESSand oneCANCELLED, and GitHub's status rollup treats the cancelled one as not-passing.Root cause
When Dependabot rebases a PR it force-pushes a new commit and updates the PR body in the same instant. That fires two
pull_request_targetevents nearly simultaneously:synchronize— from the force-pushedited— from the body updateBoth resolve to the same concurrency group (
PR Title Check-<pr number>), socancel-in-progress: truecancels the first run when the second starts. Both runs had already created check-runs with identical names, leaving a danglingCANCELLEDcheck-run alongside the successful one on the head commit.The concurrency config works as intended for superseding older commits; it only backfires when two events land on the same commit.
Fix
The check only validates the PR title and description, and neither changes on a push. So
synchronizeis both semantically unnecessary and the specific trigger racing againstedited. Trigger only onopened,edited, andreopened— the events that actually change what's being validated.🤖 Generated with AI