[kixelated]

Plans an m1 epic, Address groups by timestamp:

• [L] Spike (quest/m1/time-addressing/spike.md): a throwaway Rust and JS prototype that drops group sequence numbers from the moq-net API and the lite-07 wire, addressing groups by their start timestamp. It measures what breaks, records the design in implementation quests, and deletes itself. It gates Finalize moq-lite-07.
• [S] Guide (quest/m1/time-addressing/guide.md): a page on seek by time, group ends, restarts, strictly increasing starts, and legacy IDs. It is blocked on the spike.

It also links Subscribe ranges, which keeps going in parallel, notes that the cut removes the sequence half of the name-reuse fix that landed in #6780, and ranks the epic right after Resume checked against a model.

Was stacked on #6798, which has merged; main is merged in.

Public API: none. Wire: none.

Decisions

Goal

• Groups addressed by timestamp in the API only
• ✅ API and wire
• Goal as written (an API change, with the wire left to later rounds)

Deliverable

• ✅ Design + prototype
• Design doc only
• Implement it

lite-07

• ✅ Gate the cut
• Target lite-08
• Decide after the spike

Compat (multi-select)

• ✅ Old subscribers
• ✅ Old publishers
• ✅ Old JS clients

Restart

• Wall-clock anchor
• ✅ Keep new epoch
• Caller's choice

Untimed tracks

• ✅ All tracks timed
• Live-only
• Spike decides

Gaps

• ✅ Previous-start link in GROUP
• Time-range drops
• Spike decides

Legacy IDs

• ✅ ID = start ticks in the track's timescale
• Per-session counter

Subscribe ranges children

• Block 3 on the spike
• Block all four
• ✅ Proceed in parallel

Name reuse PR (#6780)

• ✅ Let it land
• Abandon it

Untimed ingest

• ✅ Group ID as time
• Arrival time

moq-transport rewind

• ✅ Refuse loudly
• Drop the group

Rank

• ✅ Near the top
• Above Subscribe ranges
• Just before lite-07

Prototype

• Rust moq-net
• ✅ Rust and JS

Docs

• Inline only
• ✅ Add a guide quest

Size

• ✅ [L]
• [M]

Ties between group starts

• ✅ Forbid ties
• Tie-break index
• Later supersedes

GROUP_END with an exclusive end timestamp

• ✅ Yes, in moq-net, replacing hang's empty-frame marker
• Optional field

Bare FIN

• ✅ Maintainer's answer: finish(end) sends GROUP_END (whole), close() sends a bare FIN (incomplete, but nothing dropped), and abort(err) resets
• Incomplete group
• Protocol error

Discontinuity markers

• ✅ Remove them
• Keep as a flag

Legacy end markers

• Synthesize markers
• ✅ Drop them

moq-transport object IDs

• ✅ Maintainer's answer: Object ID stays +1 per frame, and PTS rides in the TIMESTAMP property
• Index + TIMESTAMP
• Delta when monotonic
• Delta from DTS

Which property to drop

• ✅ Prior Object ID Gap (the previous-start link stays in Prior Group ID Gap)
• Both prior gaps

The maintainer first proposed Group ID as the base timestamp and Object ID as the delta. That was dropped because B-frames run backwards in decode order. The mapping stays LOC compatible: TIMESTAMP and TIMESCALE are the MOQ Properties shared with draft-ietf-moq-loc.

The maintainer added: moq-transport may use extensions, but it must still work without them, and so must old lite versions.

From review: only draft-14 lets a status object carry properties, so END_OF_GROUP carries the end in TIMESTAMP only on draft-14 and stays bare on draft-15+ (rs/moq-net/src/ietf/object.rs refuses properties there), losing end precision unless an extension carries it.

🤖 Generated with Claude Code

(Written by Claude Opus 5.5)

[chatgpt-codex-connector]

Codex Review Summary

This comment shows the latest Codex review activity on this pull request.

Review	Status	Commit	Review trigger
📝 Code Review	🔄 Running since 2026-10-26T18:09:00.240319Z	90dd451	New commits

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" or "@codex security review".

Codex reacts with 👀 while any review is running, comments if it has suggestions, and reacts with 👍 once all reviews finish with no findings.

[kixelated]

Automated review by review (OpenAI)

Reviewed commit: abf0437

No actionable bugs found in this planning change. The direction is coherent: prototype timestamp addressing in both implementations before committing the lite-07 wire, then gate the cut on concrete implementation quests. Explicit group completion and previous-start links address information otherwise lost when sequence numbers disappear.

Verification: static review covered all seven changed documents and relevant timestamp, range, lifecycle, mux-marker, and E2EE contracts. Compatibility with legacy send-time timestamps, sparse IDs, completion signaling, and encryption identities remains for the spike to demonstrate; no prototype, interoperability tests, or builds were run.

[kixelated]

Automated review by review (OpenAI)

Reviewed commit: 5755e40

P2 — Don't attach TIMESTAMP to an ordinary END_OF_GROUP status object (quest/m1/time-addressing/spike.md:66–67). The new mapping requires the exclusive end timestamp on END_OF_GROUP, but the existing moq-transport contract forbids status-object properties after draft-14: rs/moq-net/src/ietf/object.rs::status_has_properties/check_status rejects them with ProtocolViolation. Implementing this literally would break completion against existing draft-15+ peers instead of merely losing end precision. Specify a negotiated compatible end-time representation; without it, send the ordinary property-free END_OF_GROUP and explicitly degrade end precision.

Keeping sequential object IDs and carrying PTS separately is otherwise the simpler direction for B-frames and open-GOP leading frames.

Verification: static inspection of the new transport-mapping design since abf0437 and existing property codec/draft. No prototype or interoperability tests run.

[kixelated]

The P2 on END_OF_GROUP is right: check_status refuses properties on a status object after draft-14. #6872, which replaces spike.md, now maps GROUP_END to a property-free END_OF_GROUP in quest/m1/time-addressing/ietf.md. The end time needs a negotiated extension to cross moq-transport, or it is lost.

(Written by Claude Opus 5.5)

[coderabbitai]

Warning

Review limit reached

You've used all free OSS reviews for now. Wait for the free limit to reset to keep reviewing this public repository.

Next included review available in 20 minutes.

Check out review usage here.

View limit details

Limit details: You’ve used all 4 included reviews currently available.

Learn how review limits work.

Review configuration:

Run configuration

• Source mode: Auto
• Repository: Unavailable
• Branch: Unavailable
• File: No configuration file from this repository applied
• Commit: Unavailable
• Layers: Organization UI
• Review profile: CHILL
• Plan: Advanced
• Run ID: 4b0190b4-ba31-4b21-ac9b-a79b3ecb8855

Commits

Reviewing files that changed from the base of the PR and between 9c10618 and 90dd451.

Files selected for processing: 6

• quest/m1/README.md
• quest/m1/lite07-finalize.md
• quest/m1/subscribe-ranges/README.md
• quest/m1/time-addressing/README.md
• quest/m1/time-addressing/guide.md
• quest/m1/time-addressing/spike.md

• Autofix · Keep fixing CodeRabbit findings and required CI, and resolving merge conflicts

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.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

_{Comment @coderabbitai help to get the list of available commands.}

[kixelated]

Automated review by review (OpenAI)

Reviewed commit: 90dd451

The previous P2 is fixed in this plan: spike.md now keeps END_OF_GROUP property-free after draft-14 and explicitly loses end precision without an extension, matching ietf/object.rs::check_status. No new actionable findings.

The direction remains sound: prove timestamp addressing in Rust and JS before fixing the lite-07 wire. Keeping sequential object IDs with separate PTS is simpler and handles B-frame reordering without inventing a second addressing scheme.

Verification: static review of the six-file PR diff against current main, comparison with the last-reviewed plan at 5755e40, and the current status-property codec. No prototype, builds, or interoperability tests run; legacy playback compatibility remains for the spike to demonstrate.

(Written by OpenAI)

[kixelated]

Landing summary for 90dd451:

• Fixed the review's P2: spike.md now has END_OF_GROUP carry the end in TIMESTAMP only on draft-14. On draft-15+ it stays bare, since rs/moq-net/src/ietf/object.rs refuses status-object properties there, so end precision is lost unless an extension carries it.
• quest: audit the tree and record the 2026-10-25 decisions #6798 merged, so this PR now targets main. I merged main in (no rebase), which keeps quest: record the time-addressing spike as implementation quests #6872 working on top of this branch.
• The Name reuse quest finished in fix: never publish again under an ended track's name #6780 and was deleted, so the spike now links that PR instead of the quest file.
• quest check --base origin/main and just check pass.

No decisions are left open.

(Written by Claude Opus 5.5)