[kixelated]

Problem

Since moq-dev/moq#7044, a C publisher announces with moq_publication_announce, which took the shared moq_route. That struct had no epoch, so a C publisher could not announce a restart, and it accepted any caller-supplied hops, so a local publisher could claim relay hops it never took. moq-ffi's MoqAnnounceRoute already carries only a cost and an epoch. C also had no way to mint an epoch, so moq_broadcast_request_prepare needed one generated by hand.

Approach

Quest: quest/m1/c/announce-route.md (deleted here, finished).

• moq_announce_route { cost, epoch, epoch_len } mirrors MoqAnnounceRoute. moq_publication_announce takes it; NULL or zeroed is still the default route. An epoch that is not canonical UUIDv7 text fails with MOQ_ERROR_INVALID_CONFIG.
• moq_mint_epoch(char *dst, size_t dst_len) writes MOQ_EPOCH_LEN (36) bytes of fresh UUIDv7 text, like moq-ffi's moq_mint_epoch.
• moq_route now describes a received route: moq_announce_event gains a trailing route field (hops, cost, epoch), borrowed from the announcement like prefix and captures. Before this, no call returned a moq_route, so it had nothing to describe once announcing moved off it.
• Tests: a minted epoch is ordered UUIDv7 text, short or NULL buffers are refused; a locally announced route reaches a listener with its cost and epoch and no hops; a non-UUIDv7 epoch is refused; a re-announce under a new epoch reaches the listener as MOQ_ANNOUNCE_KIND_RESTART carrying the new epoch. The anonymous-hop announce test is gone along with announced hops.
• doc/lib/c gains a short Publish sample (compiled by just rs c-tests), and the rs/moq-c README lists the new signature.

Unchanged: c/moq's probe, the C++ package, and the OBS plugin already announce through moq-ffi's generated AnnounceRoute and mint_epoch.

Choices made without asking (open to review):

• I wired moq_route into moq_announce_event instead of deleting it. This keeps the planned "stays for received routes" role and matches MoqAnnounce.route. Deleting it would be smaller, but C subscribers would then have no way to see an epoch or hops.
• moq_route leaves out moq-ffi's anonymous flag, since a C caller can find it by scanning hops for a 0.
• moq_mint_epoch writes into a caller buffer sized by MOQ_EPOCH_LEN and returns the length, rather than returning a struct by value. This matches how other C inputs take const char * plus a length.
• A bad epoch maps to the existing MOQ_ERROR_INVALID_CONFIG. I did not add a dedicated code like moq-ffi's InvalidEpoch.
• No MIGRATE.md edit, since the root AGENTS.md says only the release cut writes those sections. The Impact bullet below is the migration note.

Impact

• Breaking (C ABI): moq_publication_announce(uint32_t, const moq_announce_route *) replaces const moq_route *. Migrate &(moq_route){.cost = c} to &(moq_announce_route){.cost = c} and drop any hops, which the origin now builds. To announce a restart, mint a fresh epoch per run with char e[MOQ_EPOCH_LEN]; moq_mint_epoch(e, sizeof e); and set .epoch = e, .epoch_len = sizeof e. Viewers then see MOQ_ANNOUNCE_KIND_RESTART instead of a stalled broadcast.
• Breaking (C ABI layout): moq_route loses its role as an input and gains moq_string epoch. moq_announce_event gains a trailing moq_route route. Rebuild against the new header.
• Added: moq_mint_epoch, MOQ_EPOCH_LEN, moq_announce_route.
• Wire: none.

Alternatives

• Refuse non-empty hops on the shared moq_route, or validate hop values. The maintainer rejected both on 2026-10-28 in favor of a separate struct that mirrors moq-ffi.
• Delete moq_route outright, as described above.

Follow-ups

• moq_broadcast_request_prepare reports a malformed epoch as MOQ_ERROR_INVALID_CODE, which reads as wrong. MOQ_ERROR_INVALID_CONFIG would match moq_publication_announce. This is a small break, so I left it out of this PR.
• No moq-c test covers hops on a route received over a relay session. The local tests show only that hops are empty.

Checks: just ci passed locally. So did just rs c-tests (C fixtures plus doc samples against the generated moq.h), cargo test -p moq-c (149 passed), cargo clippy -p moq-c --all-targets --all-features -D warnings, and quest check --base origin/main.

(Written by Claude Opus 5.5)

🤖 Generated with Claude Code

[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 34 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: 9738732e-60b7-487d-bb24-8c4bed93930f

Commits

Reviewing files that changed from the base of the PR and between 53c0dcd and a64b5a5.

Files selected for processing: 7

• doc/lib/c/index.md
• quest/m1/c/README.md
• quest/m1/c/announce-route.md
• rs/moq-c/README.md
• rs/moq-c/src/api.rs
• rs/moq-c/src/origin.rs
• rs/moq-c/src/test.rs

• 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.}

[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	✅ Completed 2026-10-28T22:33:30.809936Z	a64b5a5	Draft marked ready

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: a64b5a5.

No actionable bugs found in the changed code. Separating publish inputs from observed routes is the right direction: callers can supply cost/epoch without forging relay hops, matching the existing moq-ffi conversion. The fixed-length mint API reuses Epoch validation/generation, and received route pointers borrow allocations owned by the announcement record until it is freed. Keeping the received-route struct is justified by exposing restart identity and routing information. The C signature/layout break is explicitly documented.

Verification: static review of the diff, announcement ownership/lifetime, FFI string parsing, Epoch implementation, header-generation setup, and moq-ffi counterpart. I did not build the generated C header, run tests, or verify received relay-hop behavior; that integration coverage remains a documented gap.

[kixelated]

Merging at head a64b5a57b0c3b99e78ef3be98eca53d5b88e7717.

• Reviews: Codex reacted with a thumbs-up (no findings), and the automated review posted from the maintainer's account found no actionable bugs. No unresolved threads.
• CI: all 12 checks passed on this head.
• quest check --base origin/main --branch quest/m1/c/announce-route passes. The quest file is deleted here, so the quest is finished.
• No changes since review. The maintainer approved merging.
• The choices listed in the body (keep moq_route for received routes, no anonymous flag, caller-buffer moq_mint_epoch, reuse MOQ_ERROR_INVALID_CONFIG) stand as written.
• The C API gaps follow-up is planned in moq-dev/moq#7195.

(Written by Claude Opus 5.5)