Skip to content

Refresh NOTICE and fail a stale one during review - #12

Merged
cloudant42 merged 1 commit into
mainfrom
fix/stale-notice
Aug 11, 2026
Merged

Refresh NOTICE and fail a stale one during review#12
cloudant42 merged 1 commit into
mainfrom
fix/stale-notice

Conversation

@cloudant42

Copy link
Copy Markdown
Owner

The 0.1.1 server and client releases both stopped at the same gate: web/package-lock.json had moved nanoid to 3.3.18 and postcss to 8.5.23, while the committed NOTICE still named 3.3.16 and 8.5.22. Both release workflows refuse to publish a stale file, so neither could get past it.

NOTICE is regenerated here from the lockfile that is actually committed. The only content change is those two version numbers.

Why it reached main

npm-licenses regenerated NOTICE over the top of the committed one and then only inspected the result, so a pull request that moved a dependency always looked clean while the release workflows, which compare the two, would refuse to publish. Staleness was invisible until release.

That job now runs the same comparison the release does and fails when they disagree, writing the generated file to a temporary path so the committed one is still what gets inspected and uploaded. A dependency bump now fails on the pull request that causes it. This matters for the dependabot pull requests currently open, several of which touch the lockfile.

Release state

Both failed runs stopped before publishing anything: no server-v0.1.1 or client-v0.1.1 tag, no release, no draft. All four client build legs passed, including both macOS targets and the native-keyring gate. Re-running after this merges is safe.

The 0.1.1 server release stopped at its own gate: web/package-lock.json
had moved nanoid to 3.3.18 and postcss to 8.5.23, while the committed
NOTICE still named 3.3.16 and 8.5.22. The generated file is regenerated
here from the lockfile that is actually committed.

The drift reached main because npm-licenses regenerated NOTICE over the
top of the committed one and only inspected the result, so a pull request
that moved a dependency looked clean while the release workflows, which
compare the two, would refuse to publish. The job now runs the same
comparison the release does and fails when they disagree, which moves the
failure to the pull request that causes it. The generated file is written
to a temporary path so the committed one is still what gets inspected and
uploaded.
@cloudant42
cloudant42 merged commit 65a2e0e into main Aug 11, 2026
12 checks passed
@cloudant42
cloudant42 deleted the fix/stale-notice branch August 11, 2026 22:36
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