Skip to content

docs: correct two claims that no longer match reality - #33

Merged
ivanvyd merged 1 commit into
mainfrom
fix/stale-release-docs
Aug 1, 2026
Merged

docs: correct two claims that no longer match reality#33
ivanvyd merged 1 commit into
mainfrom
fix/stale-release-docs

Conversation

@ivanvyd

@ivanvyd ivanvyd commented Aug 1, 2026

Copy link
Copy Markdown
Owner

Both found by an independent release-readiness reviewer, both verified at source before acting.

RELEASING.md misdocumented the live trusted-publishing policy

The table still recorded Repository: lakespeak. The repo was renamed to LakeSpeak.NET and the live policy on nuget.org was updated to match — but that table is explicitly framed as a record of what is configured, so a stale row is worse than no row: it is precisely what someone would rebuild the policy from after it broke.

The reviewer rated this critical on the assumption the live policy was also stale, which would have failed the OIDC exchange. It isn't — the policy was updated. So this is a real documentation defect, not a release blocker.

The provenance action pin claimed the wrong version

actions/attest-build-provenance@0f67c3f… carried # v2. That commit is actually tagged v4.1.1; the real v2 is a different commit (e8998f9…).

The pin works — two rehearsal runs produced valid SLSA attestations naming both packages — so the SHA is kept and the comment corrected, rather than re-pinning to an older major just to make a comment true.

Swept the rest

Audited all 13 pinned actions against the GitHub tags API. Every other version comment matches the tag its commit actually carries, CodeQL included (its four entries share one SHA correctly tagged v4.37.3).

No functional change. The pending release run is unaffected — it executes the workflow from the tagged commit, not from main.

RELEASING.md's trusted-publishing table still recorded the policy Repository as
lakespeak. The repository was renamed to LakeSpeak.NET, and the live policy was
updated to match -- but that table is explicitly framed as a record of what is
configured, so a stale row there is worse than no row: it is the thing someone
would rebuild the policy from after it broke.

The provenance action was pinned with a comment claiming v2. The pinned commit
0f67c3f is actually tagged v4.1.1; the real v2 is a different commit entirely.
The pin itself is fine and has produced valid attestations in two rehearsal
runs, so the SHA is kept and the comment corrected rather than re-pinning to an
older major.

Audited every other pinned action while here: all thirteen version comments
match the tags their commits actually carry.
@ivanvyd
ivanvyd merged commit 3913e69 into main Aug 1, 2026
10 checks passed
@ivanvyd
ivanvyd deleted the fix/stale-release-docs branch August 1, 2026 14:29
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant