Repository navigation
quest: delete backport-audio-group-duration; @moq/publish 0.5.5 shipped - #7279
Conversation
Quest-Claim: 1c7b3859be939d0e
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
#7290 merged into release unbumped; npm's latest @moq/publish is still 0.5.4, which lacks groupDuration. Rewrite the quest to the remaining condition: the maintainer's combined cut publishes 0.5.5, then delete it. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
… npm Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
|
No actionable comments were generated in the recent review. WalkthroughThe README no longer lists the “Backport audio groupDuration” quest. The backport plan describing the proposed Merge Risk:
|
There was a problem hiding this comment.
✅ No new issues found.
Reviewed changes
- Quest deleted:
quest/m0/backport-audio-group-duration.mdis removed now that its Goal shipped, with no archive copy (the repo's convention for a completed quest). - Epic link removed: the
quest/m0/README.mdbullet for that quest is dropped, and no other file references it.
I re-verified the completion evidence rather than taking it from the body: #7290 is merged into release as d0cff5ac8b, release carries js/publish@0.5.5 (396fd3e67, merged as 5559eb337), and the published 0.5.5 tarball on npm (gitHead 5559eb337) contains groupDuration in both audio/encoder.d.ts and the encoder bundle, defaulting to Time.Milli(0). doc/lib/js/publish.md and both API reports were updated by the backport; js/publish/README.md does not document encoder props, so nothing is missing. git grep backport-audio-group-duration is empty, and quest check --base origin/main passes here (575 documents ok).
ℹ️ The recorded audio-group cap follow-up points the wrong way
The body's first follow-up asks to cap audio groups at MAX_GROUP_FRAMES / MAX_GROUP_CACHE_BYTES, but MAX_GROUP_FRAMES no longer exists in rs/moq-net, and quest/m2/unbounded-groups decides a group sheds its front at MAX_CACHE_BYTES instead of aborting. A 2.5 ms Opus frame therefore does not reach GroupTooLarge on main, so the premise is worth re-checking before the follow-up is filed as work.
Technical details
# Stale audio group cap follow-up
## Affected sites
- PR body `## Follow-ups`, first bullet
- `quest/m2/unbounded-groups/README.md:5` — `MAX_GROUP_FRAMES` is gone since #6993
- `rs/moq-net/src/model/group.rs:1572` — `GroupTooLarge` is returned only when the group cannot trim to fit
## Open questions for the human
- Is any of this still worth a quest, or does the m2 shedding work cover long audio groups?deepseek-v4.1-flash-1017 (free via Pullfrog for OSS) | 𝕏
kixelated
left a comment
There was a problem hiding this comment.
Automated review by review (OpenAI)
Reviewed head: 88f135fb0ee16589dfec648a6862afc50e3bd49a.
Findings: No actionable defects found in the complete two-file diff. It removes only quest/m0/backport-audio-group-duration.md and its matching quest/m0/README.md entry; no runtime, API, or release code changes are introduced.
Completion evidence and fix status: The cleanup is supported by the release-line evidence, rather than the package version on main:
- #7290 merged the backport into
releaseasd0cff5ac8b0955f873eb5822891fb44b87ecc070. - #8025 merged the 0.5.5 cut as
5559eb337475a30da672433b3ad3a935dbd25691. At that exact commit,js/publish/package.jsonis 0.5.5 andjs/publish/src/audio/encoder.tscontains the publicgroupDurationoption and grouping implementation. - The successful Release JS job explicitly logs
+ @moq/publish@0.5.5followed by exit code 0.
No earlier published review by this reviewer was present to reconcile. The existing oversized-audio-group follow-up described in the PR is not fixed by this documentation cleanup and should remain separate.
Direction: No changes requested for this cleanup. Removing the finished quest is consistent with the verified backport and publication evidence.
Validation limits: GitHub-only static review of the full diff, release source, merged PRs, and workflow evidence. The exact-head Quest check passed; another Check job was still running when inspected. I did not run quest context, quest check, or runtime tests locally, nor independently download the npm tarball. This is a COMMENT review, not an approval or a claim that all CI has passed.

Closes #7267
Summary
releaseon 2026-10-30 asd0cff5ac8b. It portsAudio.Encoder'sgroupDurationfrom moq-dev/moq#4910, with the default of 0.@moq/publishto 0.5.5, and Release JS published it. On 2026-11-06,npm view @moq/publish@0.5.5 versionreturns0.5.5, and the packed tarball carriesgroupDuration(inaudio/encoder.d.tsand the encoder bundle).quest rm, which also drops itsm0README entry.mainis merged in, andquest check --base origin/mainpasses.Decisions
Review
Skipped for this head (
88f135fb0e). The diff only deletes a finished quest and its README link;quest check --base origin/mainvalidates it. The code's review is recorded on moq-dev/moq#7290 (Codexgpt-6.1-sol, medium effort, headbcd6040fd6: one finding refuted as out of scope, then a clean round).Follow-ups
main, cap audio groups atMAX_GROUP_FRAMES/MAX_GROUP_CACHE_BYTES. AgroupDurationof tens of seconds with 2.5 ms Opus frames throwsGroupTooLarge. Fix it there, then backport.releasebump. Consider having them name "the next release" instead.🤖 Generated with Claude Code
(Written by Claude Opus 5.5)