chore(release): v0.37.0 #675
No reviewers
Labels
No labels
bump
major
bump
minor
bump
patch
kind/bug
kind/chore
kind/docs
kind/feature
priority/critical
priority/high
priority/low
priority/medium
size/L
size/M
size/S
size/XL
No milestone
No project
No assignees
2 participants
Notifications
Due date
No due date set.
Dependencies
No dependencies set.
Reference
frankenbit/release-toolkit!675
Loading…
Add table
Add a link
Reference in a new issue
No description provided.
Delete branch "release-prep/rolling"
Deleting a branch is permanent. Although the deleted branch may continue to exist for a short time before it actually gets removed, it CANNOT be undone in most cases. Continue?
Added
None.
Changed
changelog-body-check gate: now runs
rt changelog-body-checkinstead of the bash script (#607)Gate 3 of 4, and the last PR-CI validator on bash. It uses the same two-arm rt bootstrap as the other three gates, so adopters pinning a release tag fetch the published binary. The five
CHANGELOG_BODY_CHECK_*inputs are unchanged and the Go port honours them.Fixed
release: the cut log now says whether the manifest was updated or created (#673)
The manifest write forks on whether the file is present in the checkout. Neither branch logged anything, so a legitimate first cut and a lost manifest looked identical. The create branch now states the assumption it is making, so a project with prior releases taking that path is visible in the log.
release: the post-cut manifest check can run again — it had no credentials (#673)
The check added in #637 fetches the default branch to confirm the bookkeeping commit landed. It ran three lines after the push credential was deleted, so it exited 128 and killed the step before the comparison. The guard reddened whether or not the commit landed, so it could not tell the two apart. It now scopes a credential to its own fetch.
reusable-release: the create echo names the ASSUMPTION, not the key count (#673)
pr-ci: gate 3 was missing the version-tag cache restore (@lookout #676)
pr-ci: make the changelog-body-check gate re-run when the gate itself changes
Removed
None.
Deprecated
None.
Upgrade
None.
chore(release): v0.36.1to chore(release): v0.37.02f62bfc0ab875a3c61d7875a3c61d78bfbbc3030APPROVED — reviewed at head
8bfbbc30, and I am naming the SHA in the body because the row binds to whatever is current at submit time, not to what I read.Gate state, re-measured rather than taken from the dispatch
The check the diff cannot do — a MISSED pin
A release cut's characteristic defect is a pin left behind, and that is invisible in a diff of what changed. Scanned the whole tree at head:
That one is correct and must not be bumped here. The manifest records the last released tag, and
reusable-release.ymlwrites.last_released_tag=$tagat cut time — so it is stillv0.36.0until this cut runs and pushes the bookkeeping commit. Changing it in the PR would make the manifest claim a release that has not happened.CHANGELOG fidelity — and a count that does not match on its face
Three fragments were consumed and six entries produced, which is worth stating because it looks wrong:
Not a defect:
scripts/release-prep.sh:277-278names both sources — "no fragments in changelog.d/ AND no conventional commits since : nothing to release … add fragments or write feat:/fix: commits". The two#673entries are distinct changes (the branch marking, and the later echo refinement), not a duplication.changelog.d/is empty at head. Version0.36.0 → 0.37.0; a minor bump is right for a batch containing a behaviour change (gate 3 moving tort). Date and README status line both2026-08-17.What this review does NOT cover
reusable-release.ymlonly far enough to confirm it writes the manifest.Content is a clean release cut. Merging is @bosun's gate and I am not taking it.