Repository navigation
fix(moq-net): reject incompatible copies during failover - #3521
Conversation
Codex Review SummaryThis comment shows the latest Codex review activity on this pull request.
ℹ️ About Codex in GitHubYour team has set up Codex to review pull requests in this repo. Reviews are triggered when you
Codex reacts with 👀 while any review is running, comments if it has suggestions, and reacts with 👍 once all reviews finish with no findings. |
|
No actionable comments were generated in the recent review. 🎉 ℹ️ Recent review info⚙️ Run configurationConfiguration used: Organization UI Review profile: CHILL Plan: Advanced Run ID: 📒 Files selected for processing (1)
🚧 Files skipped from review as they are similar to previous changes (1)
Included review availability: Your plan provides up to 4 included reviews per hour; 3 remain after this review. WalkthroughThe origin front now records immutable track metadata for each track name. During source splicing, incompatible metadata is rejected with Severity of issue fixed: Medium Merge Risk: ⚪ Minimal · up to The change rejects incompatible failover sources while keeping compatible routes eligible, with regression coverage described for the affected behaviors. No concrete merge-blocking risk remains. 🚥 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 |
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: 7c7fc9797e
ℹ️ About Codex in GitHub
Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you
- Open a pull request for review
- Mark a draft as ready
- Comment "@codex review".
If Codex has suggestions, it will comment; otherwise it will react with 👍.
Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".
7c7fc97 to
e350980
Compare
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: e350980842
ℹ️ About Codex in GitHub
Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you
- Open a pull request for review
- Mark a draft as ready
- Comment "@codex review".
If Codex has suggestions, it will comment; otherwise it will react with 👍.
Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
`resume::Consumer::poll_info` resolved `track::Info` from segment zero while every data read routes to the newest segment. When a track is re-served on a live path (a source dies and a standby takes over, or a publisher reconnects), those are different generations, so a subscriber was handed the predecessor's metadata alongside the successor's frames. `timescale` is the damaging field. On moq-lite the model converts each frame timestamp into its own track's timescale and the raw value goes straight onto the wire, so TRACK_INFO advertising the predecessor's scale silently rescales every frame rather than failing. `priority` and `latency_max` were wrong but survivable. Resolve from the newest segment instead, and park on the segment list while its info is still pending so a takeover that supersedes it re-resolves against the successor rather than answering with a generation nobody will read. This is the Rust half of the defect #3422 fixed in `js/net`. No wire or API change, so no draft moves. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Co-Authored-By: GPT-6 <noreply@openai.com>
Preserve immutable metadata for each broadcast track and reject incompatible routes before splicing. Keep healthy copies eligible and retain the contract across idle release and reopening. Co-Authored-By: GPT-6 <noreply@openai.com>
e350980 to
64ccc33
Compare
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: 64ccc3317e
ℹ️ About Codex in GitHub
Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you
- Open a pull request for review
- Mark a draft as ready
- Comment "@codex review".
If Codex has suggestions, it will comment; otherwise it will react with 👍.
Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".
| _ => match state.write() { | ||
| Ok(mut state) => state.accept_track_info(&name, info).map(|()| track), | ||
| Err(_) => Err(Error::Dropped), |
There was a problem hiding this comment.
Revalidate the route before caching its metadata
When a better route attaches while the previous route's TRACK_INFO is pending, both the query and source-table poll can become ready, but the wait closure checks the query first and reaches this write without confirming that serve_route(skip) is still id. This can cache the superseded route's metadata, causing the preferred route to be rejected as Unsupported if its properties differ and leaving the track on the obsolete route or aborting when that route disappears. Fresh evidence beyond the earlier resume race is this new FrontState cache write after the table-aware wait; revalidate the selected route under the write lock before accepting its info. rs/moq-net/CLAUDE.mdL10-L10
Useful? React with 👍 / 👎.
Resolved by behavior, keeping both sides' fixes: - moq-net ietf publisher: main's joining FETCH with its saved object prefix (#3562) rebuilt on dev's shapes; its two refusal codes map to DOES_NOT_EXIST until the registry gains them (quest/m2/ietf-leftovers.md) - moq-net origin: main's incompatible-copy refusal during failover (#3521) ported to dev's source table (FrontState::track_info, checked on Splice) - draft-21 (#3574): the moq-tokio client/server/websocket arms, the CLI version choice lists, the IETF error registry gates and test lists, and js/net accept.ts, which the auto-merge had left referencing a removed local - moq-video/moq-audio: main's Start::Latest and drain fixes (#3565, #3568) expressed with dev's max_age; ported tests carry an age budget - quest/: dev's tree with main's decisions since the merge base (#3576, #3619, #3623, #3604), quests finished on either side deleted Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
Summary
Unsupportedand keep compatible routes eligible.quest/m0/resume-info-newest.md.Public API
None. All added state and validation are private.
Wire
No format changes. Incompatible failover sources are refused using the existing
Unsupportederror.(written by GPT-6)