ci: drop --order-dependents from the release publish trigger - #1790
Conversation
melos builds its ordering graph from dev_dependencies too, so supabase_testing (which depends on supabase) plus the client packages that dev-depend on supabase_testing form a cycle, and melos exec exits before dispatching anything.
|
No actionable comments were generated in the recent review. 🎉 ℹ️ Recent review info⚙️ Run configurationConfiguration used: Path: .coderabbit.yaml Review profile: CHILL Plan: Team Run ID: 📒 Files selected for processing (1)
Included review availability: Your plan provides up to 4 included reviews per hour; 3 remain after this review. 📝 WalkthroughWalkthroughThe release-tag workflow now runs ChangesRelease workflow
Estimated code review effort: 2 (Simple) | ~5 minutes Merge Risk: ⚪ Minimal · up to This localized CI change removes a failing dependency-ordering flag so release publish workflows can be dispatched despite the workspace cycle; no actionable merge-blocking risk remains beyond normal checks and review. Suggested reviewers: 🚥 Pre-merge checks | ✅ 5✅ Passed checks (5 passed)
Full details: Docstring CoverageExplanation No functions found in the changed files to evaluate docstring coverage. Skipping docstring coverage check. Docstring coverage is scoped to functions touched by this diff. Analyzed 0 functions across 0 files. (1 skipped: 1 unsupported.) ✨ Finishing Touches🧪 Generate unit tests (beta)
Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out. Comment |
What kind of change does this PR introduce?
CI fix. The
Create Release Tagsworkflow failed on the v3.0.0-dev.2 release (run 33512894194): the tags were created and pushed, but theTrigger publish workflowsstep exited before dispatching anything, so nothing reached pub.dev.What is the current behavior?
melos exec --order-dependentsbuilds its ordering graph fromdev_dependenciesas well asdependencies. Since #1747,supabase_testingdepends onsupabase, and nine packages dev-depend onsupabase_testing, which closes a cycle throughpostgrest. Melos checks for cycles across the whole workspace before running anything and exits 1 when it finds one, so the flag cannot be combined with this dependency graph. Filtering the cycle away with--ignoredoes not help either, because the check runs over all packages rather than the filtered set.This is the first run of the workflow since
supabase_testinglanded. The previous release commit was titledchore: publish packages as 3.0.0-dev.1, which does not match the workflow'schore(release):condition, so the job was skipped; the last run before that was on 2026-08-05, before the package existed.What is the new behavior?
The flag is dropped, with a comment recording why it cannot come back. Dispatch order does not affect the outcome: every tag points at the same release commit, each dispatched
Publish Packagesrun publishes whatever is unpublished at that commit, andmelos publishitself runs unordered at concurrency 1 and resolves each package against the workspace rather than against pub.dev.Verified locally against
main: the command as it stands reproduces the cycle error, and without the flag it enumerates all 12 unpublished non-private packages with the expected tag refs.Additional context
This does not retrigger the stuck release.
Create Release Tagsruns on push tomain, and the tags for3.0.0-dev.2already exist, so publishing that release still needs a manualPublish Packagesdispatch on one of those tags.Summary by CodeRabbit