Repository navigation
chore: release v0.9.40 - #7617
Conversation
|
Preview deployment for your docs. Learn more about Mintlify Previews.
💡 Tip: Enable Automations to automatically generate PRs for you. |
There was a problem hiding this comment.
Verdict: REQUEST CHANGES · 1 blocker · 0 important · 0 nits · head e306e6f5
- blocker — the release is drafted on a stale base, so what merges is not what was noted or walked. The parent is
54bcdc616(#7609).mainis 4 commits ahead: #7615 (cli add --dir), #7503 (test), #7181 (engine: frame pixels independent of prior frames,browserManager.ts) and #7611 (Node 22.23.5 in the Dockerfiles).publish.ymlbuilds frommerge_commit_sha, so v0.9.40 would ship #7181/#7615/#7611 with no entry inreleases/v0.9.40.mdordocs/changelog.mdx. The Desktop walk didn't cover them either. The PR-body claim "one commit on top of currentmain" doesn't hold.Release checkspassed only becauseHEAD^1is the stale parent. Inmerge_group,release-only-diff.mjs --verifywill list these as unlisted commits and eject the PR. Fix: rebase onmain, rerunbun run release:prepare 0.9.40, and redo the Desktop walk with the rebuilt packs (#7181 changes engine rendering).
Scope
Audited the full diff: 20 manifests, all 0.9.39 → 0.9.40, no other version strings; notes and changelog link all 5 commits in v0.9.39..54bcdc616. Also checked scripts/ci/release-only-diff.mjs, the ci.yml release-checks job and the publish.yml trigger. All Actions runs at this head pass. This is a release PR, not a fix PR, so the "fails without the fix" line doesn't apply.
e306e6f to
68c0525
Compare
jrusso1020
left a comment
There was a problem hiding this comment.
Reviewed at 68c0525ab85474b9e3f0d813f8f27860b3302c04, after the head moved from e306e6f5 during review. No blocker found at this revision.
The stale-release issue raised by @heygengenesis is resolved: the parent is 5bf361f52, the fetched main tip, and both notes surfaces now link all nine commits since v0.9.39, including the four absent from the earlier head.
Audited: all 20 manifest diffs (only 0.9.39 → 0.9.40, other JSON fields equal), the complete new release note and matching changelog entry, release baseline/ancestry, release verifier, and publisher's immutable merge-SHA checkout. The actual PR diff is 22 files, versions and notes, with no runtime changes. This is release packaging, so the lane's fix-test line does not apply.
Validation through pytest-gate.sh: node scripts/ci/release-only-diff.mjs --verify passes at this head (release only: 22 files, versions and notes). Independent manifest and commit-link checks agree. I did not build or inspect packed artifacts, rerun the underlying feature suites, or repeat the Desktop walk; this verdict covers the release diff and consistency checks.
Need: yes, publishes the accumulated changes · Size: ok, one release · Reuse: existing release preparation/verifier and merge-SHA publisher · Simpler: already versions and notes only · Structure: fine (checked package/plugin manifest conventions and both generated notes consumers; no new helpers or separable runtime changes).
Risk: medium — the release publishes multiple packages, although this diff only changes versions and notes.
Verdict: APPROVE
Reasoning: Versions agree and the regenerated notes now cover the code that the current release base will ship.
— Rames
There was a problem hiding this comment.
Verdict: REQUEST CHANGES · 1 blocker · 0 important · 0 nits · e306e6f5 → 68c0525a
- blocker — the base is stale again. The recut fixes the earlier gap: the notes now link all 9 commits in
v0.9.39..5bf361f52. Butmainhas since gained97df93fb0, #7618 (fix(cli): WAV byte-rate handling incheckAudio.ts,engine/utils/ffprobe.ts,parsers/mediaStreams.ts). That's shipping code. It isn't inreleases/v0.9.40.md, andskipsChangelogdoesn't exempt it. So the merge commit thatpublish.ymlbuilds would ship it unlisted, and themerge_grouprun ofrelease-only-diff.mjs --verifywill eject the PR. Fix: rebase, rerunbun run release:prepare 0.9.40, and queue right away somaincan't move again before the queue run.
Scope
The diff is release-only: 22 files, 20 manifests all at 0.9.39 → 0.9.40, plus the notes and changelog. All Actions runs at this head pass. This is a release PR, so the "fails without the fix" line doesn't apply.
68c0525 to
0a2494a
Compare
There was a problem hiding this comment.
Verdict: APPROVE · 1 resolved, 0 open, 0 new · 68c0525a → 0a2494af
Scope
The base is current main (33304cff0, nothing newer). The notes and changelog link all 11 commits in v0.9.39..33304cff0, including #7618 and #7606. The diff is release-only: 20 manifests at 0.9.39 → 0.9.40, plus the notes and changelog. Every Actions run at this head passes, including Release checks.
What
Release v0.9.40: version bump,
releases/v0.9.40.md, and the docs changelog. Merging publishes to npm.Why
Ships the fix for local exports that filled the disk: a render now extracts each video at about the size it is shown, so 4K footage in a 1080p film no longer expands to full 4K frames (#7609, #7607). It also carries every other change on
mainsince v0.9.39: a frame's pixels no longer depend on the frames before it (#7181),hyperframes checkhearing a WAV whose header misstates its byte rate (#7618), render work sized to the CPUs the process may use (#7612), Node 22.23.5 in the Docker images (#7611),hyperframes add --dirnaming the init command for a missing folder (#7615), the runtime element handles for the editing channel (#7603) and thehyperframes registry searchfix (#7610).What people see
No visible change
This PR is the version bump and changelog only. Each change above was captured in its own PR.
Test plan
bun run release:prepare 0.9.40ran onmainat its current tip with all tags fetched; the notes list 11 entries from v0.9.39 and the compare link readsv0.9.39...v0.9.40.node scripts/ci/release-only-diff.mjs --verifypasses on this commit, including the check that the notes link every commit since the last release.maintip it was cut from, and no tag was pushed.