fix: generalize the dependabot first-party @v1 exemption comment - #428
Conversation
The header comment claimed first-party refs and examples 'track the @v1 major tag', but capabilities now pin various major tags -- @v2 for 20+ of them (preview, claude-code-review, ...), @v1 for the rest. The blanket @v1 made the written rationale for the Dependabot exemption read as stale/wrong. Reword to name the tag generically ('their capability's major tag', pointing at README's Versioning section) so it cannot go stale again, and drop the specific @v1 from the Morrison-Lab/gha/* glob. Scope: CHANGELOG.md:328's near-identical wording (d-morrison/gha/*@v1, #48) is a released historical entry and is left as-is, per the org-move fragment's 'historical entries keep the name/state they had when written' convention -- it accurately records what was true at #48's time. Closes #310 Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
This comment has been minimized.
This comment has been minimized.
This comment has been minimized.
This comment has been minimized.
…1-doc fixes) The repo consistently adds a .fixed.md fragment for this class of fix (pin-check-news-v1-to-v2, fix-remaining-stale-v1-examples, retarget-doc-urls-after-org-move). Non-blocking, but flagged for consistency. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
ARD summaryReview verdict was Ready for merge with one optional (non-blocking) note. Addressed in
The reviewer independently re-ran the whole-repo grep and confirmed the sweep is complete: the only other hit ( |
|
Claude finished review — View run The only change since the last review round is a new commit (
This fully resolves the only outstanding item (the optional changelog-fragment nit) from the prior review, which found no blocking issues and reached a Ready for merge verdict. No new findings. VerdictReady for merge |
Closes #310
The bug
.github/dependabot.yml's header comment stated that first-partyMorrison-Lab/gha/*@v1self-references and theexamples/templates"track the
@v1major tag". That blanket@v1is stale: capabilities now pinvarious major tags --
@v2for 20+ of them (preview,claude-code-review,update-snapshots, ...),@v1for the rest. The comment is the writtenrationale for why first-party refs are exempt from Dependabot, so a reader
checking whether the exemption still applies got a tag number that no longer
matches what the repo pins.
The fix
Name the tag generically -- "their capability's major tag (see the Versioning
section of README.md)" -- so the rationale cannot go stale again, and drop the
specific
@v1from theMorrison-Lab/gha/*glob. The policy itself (first-partyrefs track the major tag rather than being SHA-pinned) is unchanged and still
correct.
Scope note
@v1is not universally stale -- it is still the correct pin for manycapabilities -- so this is deliberately not a blanket
@v1->@v2sweep; onlythe one comment that made a blanket
@v1claim is touched.A grep for the same rationale restated elsewhere found one hit:
CHANGELOG.md:328-329(d-morrison/gha/*@v1... "still track the@v1majortag", #48). That is a released historical changelog entry and is left as-is,
per the org-move fragment's own convention that historical
CHANGELOG.md/changelog.d/entries keep the name and state they had when written -- itaccurately records what was true at #48's time.