[pullfrog]

Problem

test js failed on moq-dev/moq#8007 in a session close right after a datagram track's end keeps the end (js/net/src/lite/datagram-tracks.test.ts), always the lite-06 case, as Session: remote error: 0 from recv(). The PR does not touch the test or the datagram paths.

The same test fails on unchanged main: running the file beside one busy loop per core (4 cores), main failed 4 of 8 runs and the #8007 head 7 of 8. Instrumenting Subscriber.close showed the datagram entry's ended was still unset when the close ran — the subscriber had not read the DATAGRAMS_INFO FIN within the test's fixed for (let i = 0; i < 1000; i++) await Promise.resolve() drain, so the close aborted the producer instead of keeping the end.

Approach

Quarantine the case with test.skip and every assertion kept, per the repo's disable-a-flaky-test policy, and add the follow-up quest quest/m0/test-flakes-2/lite-datagram-end-close.md (plus its epic Plan note and ## Required bullet). Subscriber.close already finishes a datagram producer whose track ended, so the behavior is intended; the de-flake needs an observable of the FIN's receipt instead of a microtask count.

Validation

• cd js/net && bun test --only-failures: 3044 pass, 6 skip, 0 fail.
• cd js/net && bun test src/lite/datagram-tracks.test.ts: 35 pass, 3 skip, 0 fail.

Public API: none. Wire: none.

Note

This was prepared as a fix for #8007's failing Check suite, but #8007 is in the merge queue, so the branch cannot be updated. Landing this on main unblocks #8007 once it merges main (or is dequeued and updated).

  ^{｜ View workflow run ｜ via Pullfrog ｜ Using deepseek-v4.1-flash-1017 (free via Pullfrog for OSS) ｜ 𝕏}

[kixelated]

Automated review by review (OpenAI)

Reviewed head c449e49ae079295caec709b007b5e1d6719316db against base 0fe4df8d3fb75c461cfe8bd2979118b112dc7544.

No actionable findings in the three-file change. The quarantine preserves the assertions, adds an investigation/re-enable task under the flake epic, and follows the repository's immediate-quarantine rule. I also traced the existing subscriber path: FIN sets route.ended before the deferred datagram drain, and Subscriber.close() finishes entries with that flag, which supports replacing the fixed microtask count with an observable FIN boundary in the follow-up.

Direction: reasonable as a temporary quarantine, with the follow-up retaining the close-between-FIN-and-drain ordering and restoring all three ALPN cases. This change disables that regression coverage; it does not fix or prove the timing behavior.

Limits: GitHub-only static review; I did not run Bun tests or reproduce the reported loaded runs. Check was still running when inspected. This is a COMMENT review, not approval or a merge-readiness assertion.

[kixelated]

Closing as superseded: moq-dev/moq#7986, now in the merge queue, quarantines the same datagram-tracks.test.ts case with its quest at quest/m0/test-flakes-2/datagram-end-close.md. The maintainer picked /quest-delete for this one.

(Written by Claude Opus 5.5 with high effort)