Repository navigation
quest: drop the JS qmux clean-finish check - #4957
Conversation
Co-Authored-By: Grok 4.7 <noreply@x.ai>
Co-Authored-By: Grok 4.7 <noreply@x.ai>
|
Outcome: the bug is not in (Written by Grok 4.7) |
|
No actionable comments were generated in the recent review. 🎉 ℹ️ Recent review info⚙️ Run configuration
📒 Files selected for processing (3)
💤 Files with no reviewable changes (3)
Included review availability: This review used your included allowance. Your plan provides up to 4 included reviews per hour; 3 remain after this review. WalkthroughThe JS qmux finish quest document was deleted. Its entries were also removed from the Required lists in the transport-upgrade README and JavaScript quest document. No code changes are described. Priority: ⬇️ Low Merge Risk: ⚪ Minimal · up to This documentation cleanup appears mergeable after normal checks. 🚥 Pre-merge checks | ✅ 5✅ Passed checks (5 passed)
✨ Finishing Touches✨ Simplify code
Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out. Comment |
Automated review: quest: drop the JS qmux clean-finish checkDocs-only. It deletes I spot-checked the conclusion against the head source:
No issues found. One small point: the next person to touch the GOAWAY path won't find this verification in the repo. It survives only in the PR description, so the CI: green (Check, Test, Quest pass; Auto-merge and Land back-merge skipped). Verdict: MERGE (reviewed head This is an automated review, not the maintainer's decision |
|
Merging per the maintainer's
(Written by Claude Opus 5.5) |
Problem
The Rust qmux crate reported every finished send stream as
connection closed(fixed in moq-dev/web-transport#402). This quest asked whether the TypeScript peer does the same, and whether@moq/netwould treat a delivered GOAWAY or SETUP as a failure over the WebSocket fallback.Approach
It does not.
@moq/qmux0.3.3 (whatjs/netdepends on; npmgitHead9c70649, and the same close path as currentweb-transportmain) fulfills bothwriter.close()andwriter.closedafter a clean FIN. The sink awaits the FIN onto the socket, then returns. A peer STOP_SENDING still errors the writable.Checked against the published package: the existing "sequential stream writes and FIN" test, plus a uni stream closed with no payload and a bidi stream closed after one write. All three fulfill
writer.closedand putfin: trueon the wire.@moq/netwould not treat SETUP or GOAWAY as that failure today even ifclosedrejected. Lite#sendSetupwrites the message and callsWriter.close(), which swallows a close rejection and does not awaitclosed. Nothing injs/netsends a GOAWAY yet. Paths that do awaitwriter.closedafter a clean finish (announce withdrawal, SUBSCRIBE_NAMESPACE, PUBLISH_NAMESPACE) would see a fulfillment, not an error.No release and no
@moq/qmuxbump. The quest is deleted, and its links are removed from the transport-upgrade line and the JavaScript upgrade quest.Impact
Alternatives
moq-dev/web-transportthat assertswriter.closedfulfills. The existing FIN test already awaitswriter.close(), which settles the same way, and the published package does not have the bug. Left it.@moq/netto awaitclosedon SETUP. Out of scope; there is no failure to fix, and sending GOAWAY belongs to the JavaScript upgrade quest.Follow-ups
None. The JavaScript upgrade can await the GOAWAY stream's
closedon the WebSocket fallback;@moq/qmuxalready reports that as a clean close.(Written by Grok 4.7)