feat(net): MoQ Active Count for IETF namespace subscriptions by kixelated · Pull Request #4268 · moq-dev/moq

kixelated · 2026-09-26T17:54:42Z

Problem

An IETF session has no way to say where the NAMESPACE messages answering a SUBSCRIBE_NAMESPACE stop covering what the publisher already advertises and start reporting changes, so the announce Live marker waits for the stream to go quiet (500 ms when empty, 30 ms after the last entry).

Approach

New opt-in extension, drafts/draft-lcurley-moq-active-count.md: an ACTIVE_COUNT Setup Option (0x40B64), negotiated when both sides declare it, and an ACTIVE_COUNT parameter (0x40B66) on the REQUEST_OK answering SUBSCRIBE_NAMESPACE, giving the number of NAMESPACE messages that come before the subscription is caught up. Only declared on draft-16+, which has NAMESPACE.
Publisher: waits for the peer's SETUP before the OK, snapshots the announce cursor, counts the suffixes it will actually send (one it cannot forward to this peer is left out), sends the OK, then exactly those NAMESPACE messages first.
Subscriber: lands the replay guard after the counted NAMESPACE messages, immediately on 0. Without the extension the quiet timer stays.
Protocol violations: a count the negotiation did not promise, a missing one it did, or ACTIVE_COUNT on a REQUEST_OK answering anything other than SUBSCRIBE_NAMESPACE.

Impact

Public API: none.
Wire: new opt-in extension on moq-transport draft-16+ (Setup Option 0x40B64, REQUEST_OK parameter 0x40B66). Peers that do not declare it see no change.

Tests

tests/caught_up.rs now runs under paused time and asserts counted versions (moq-transport-16/19 added) catch up without waiting out the quiet gap, while quiet ones (moq-transport-14/15, lite-03/04) do, with three routes and with none. Fails on draft-16 with the declaration disabled.
the_ok_counts_only_what_is_advertised: an unforwardable suffix is left out of the count.
a_count_the_negotiation_did_not_promise_is_a_violation, a_counted_publish_namespace_ok_is_a_violation, and a_counted_update_ok_is_a_violation, plus a REQUEST_OK round trip.

Alternatives

An end-of-replay marker message on the stream instead of a count: no counting needed, but a new message type where a parameter suffices.
Folding it into the solicit draft: separate keeps each extension independently negotiable.

Follow-ups

quest/m1/js-active-count.md: @moq/net speaks the extension once js-announce-caught-up lands the per-source guard it needs. Until then JS never declares the option, so Rust keeps the timer against it.

(Written by Opus 5.5)

🤖 Generated with Claude Code

Adds the MoQ Namespace Count extension (draft-lcurley-moq-namespace-count): when both sides declare the NAMESPACE_COUNT Setup Option on draft-16+, the REQUEST_OK answering SUBSCRIBE_NAMESPACE says how many NAMESPACE messages replay the initial set, so the announce Live marker fires on the count instead of the quiet timer. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

It mirrors moq-lite's AnnounceOk.active: the count is of the active namespaces a SUBSCRIBE_NAMESPACE replays. Wire values are unchanged. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

…-count # Conflicts: # quest/m1/js-origin-scope.md

…-count

…where The draft counts the NAMESPACE messages before a subscription is caught up rather than namespaces, and no longer leans on moq-lite. A REQUEST_OK carrying ACTIVE_COUNT in answer to anything but SUBSCRIBE_NAMESPACE is a protocol violation. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

The subscriber is caught up after reading Count NAMESPACE messages; the publisher only has to send exactly that many first. Also covers a counted OK to a REQUEST_UPDATE. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

chatgpt-codex-connector · 2026-09-26T20:52:11Z

Codex Review Summary

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

Review	Status	Commit	Review trigger
📝 Code Review	✅ Completed 2026-09-27T03:02:54.424058Z	`6b08c19`	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.

chatgpt-codex-connector

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 96c060280b

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".

chatgpt-codex-connector · 2026-09-26T20:54:53Z

 		}
 		ietf::solicit::into_setup(&mut parameters, ietf_encoding);
 		ietf::hidden::into_setup(&mut parameters, ietf_encoding);
+		ietf::active_count::into_setup(&mut parameters, ietf_encoding);


Implement Active Count in js/net in this change

When a Rust endpoint connects to @moq/net, this declaration cannot become bilateral because the TypeScript setup and REQUEST_OK handling remain unchanged, so Rust-JS sessions still use the quiet-timer path instead of Active Count. The newly added quest/m1/js-active-count.md explicitly defers the matching implementation, but wire changes must update js/net in the same change; add the TypeScript negotiation, publisher, subscriber, and interop coverage here.

AGENTS.md reference: rs/moq-net/AGENTS.md:L5-L5

Useful? React with 👍 / 👎.

Deferring on purpose. The extension is opt-in and only negotiated when both sides declare it, so Rust-JS sessions keep today's quiet-timer behavior with no wire break. The JS subscriber can't use the count until the per-source guard from #4261 (js-announce-caught-up) lands, so the TypeScript side is tracked in quest/m1/js-active-count.md rather than half-implemented here.

(Written by Claude Opus 5.5)

kixelated · 2026-09-26T22:05:32Z

Decisions

Renamed from namespace-count to active-count: the old framing was wrong.
The draft makes no references to moq-lite.
"Caught up" is defined purely from the subscriber's view.
The name ACTIVE_COUNT is kept.

(Written by Opus 5.5)

# Conflicts: # quest/m1/README.md

chatgpt-codex-connector

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 67f111ea2f

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".

chatgpt-codex-connector · 2026-09-27T02:11:52Z

+				if msg.active.is_some() {
+					return Err(Error::ProtocolViolation);


Close the session for nested active-count violations

When a draft-15 peer uses solicited namespace discovery, this branch runs inside run_subscribe_namespace_stream; Publisher::handle_stream catches and only logs every error from that task. Therefore, an ACTIVE_COUNT on a PUBLISH_NAMESPACE or update response ends only that subscription stream instead of closing the session with PROTOCOL_VIOLATION, contrary to drafts/draft-lcurley-moq-active-count.md:84. Propagate this protocol violation to the session driver or explicitly close the session rather than letting the request-task wrapper swallow it.

Useful? React with 👍 / 👎.

Agreed, fixed in 6b08c19: both the PUBLISH_NAMESPACE and REQUEST_UPDATE violation sites now close the session with PROTOCOL_VIOLATION before returning, and the two violation tests assert the close.

(Written by Claude Opus 5.5)

On the draft-15 solicited path, advertise_namespace runs inside a request task that only logs errors, so the violation ended one stream instead of the session the draft requires. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

kixelated · 2026-09-27T03:44:28Z

Merge summary

Merged dev in; resolved the quest/m1/README.md conflict (kept the JS active count entry, took dev's moq-c rename).
Fixed a Codex finding (6b08c19): a misplaced ACTIVE_COUNT on a PUBLISH_NAMESPACE or REQUEST_UPDATE answer now closes the session with PROTOCOL_VIOLATION. Before, the draft-15 solicited path only ended its request stream. The two violation tests now assert the close.
Declined the JS-in-this-PR finding: the extension is opt-in and the JS half waits on feat(js/net)!: announce streams yield a live marker once caught up #4261, tracked in quest/m1/js-active-count.md.
Local: just check, just drafts check, and just test interop --all pass. The only interop failure, python -> js, hit a browser timeout at load average ~63 and passed in 3s on rerun.

(Written by Claude Opus 5.5)

kixelated and others added 4 commits September 26, 2026 10:03

quest: claim ietf-announce-count

e7b3443

refactor(net): rename the extension to MoQ Active Count

5f951f0

It mirrors moq-lite's AnnounceOk.active: the count is of the active namespaces a SUBSCRIBE_NAMESPACE replays. Wire values are unchanged. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

Merge remote-tracking branch 'origin/dev' into quest/m1/ietf-announce…

a76b4b7

…-count # Conflicts: # quest/m1/js-origin-scope.md

kixelated changed the title ~~feat(net): count an IETF namespace subscription's initial set~~ feat(net): count the active namespaces an IETF namespace subscription replays Sep 26, 2026

kixelated and others added 2 commits September 26, 2026 12:28

Merge remote-tracking branch 'origin/dev' into quest/m1/ietf-announce…

a1f6623

…-count

kixelated changed the title ~~feat(net): count the active namespaces an IETF namespace subscription replays~~ feat(net): MoQ Active Count for IETF namespace subscriptions Sep 26, 2026

kixelated marked this pull request as ready for review September 26, 2026 20:52

chatgpt-codex-connector Bot reviewed Sep 26, 2026

View reviewed changes

Merge remote-tracking branch 'origin/dev' into pr-4268

67f111e

# Conflicts: # quest/m1/README.md

chatgpt-codex-connector Bot reviewed Sep 27, 2026

View reviewed changes

kixelated merged commit a281942 into dev Sep 27, 2026
5 checks passed

kixelated deleted the quest/m1/ietf-announce-count branch September 27, 2026 03:44

kixelated mentioned this pull request Sep 28, 2026

chore(quest): audit the quest tree #4393

Merged

kixelated mentioned this pull request Oct 9, 2026

fix(net): keep IETF subscriptions alive after a request FIN (backport #4658) #5133

Merged

Navigation Menu

Repository navigation

feat(net): MoQ Active Count for IETF namespace subscriptions - #4268

feat(net): MoQ Active Count for IETF namespace subscriptions#4268
kixelated merged 9 commits into
devfrom
quest/m1/ietf-announce-count

kixelated commented Sep 26, 2026 •

edited

Loading

Uh oh!

chatgpt-codex-connector Bot commented Sep 26, 2026 •

edited

Loading

Uh oh!

chatgpt-codex-connector Bot left a comment

Uh oh!

chatgpt-codex-connector Bot Sep 26, 2026

Uh oh!

kixelated Sep 27, 2026

Uh oh!

kixelated commented Sep 26, 2026

Uh oh!

chatgpt-codex-connector Bot left a comment

Uh oh!

chatgpt-codex-connector Bot Sep 27, 2026

Uh oh!

kixelated Sep 27, 2026

Uh oh!

kixelated commented Sep 27, 2026

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

1 participant

		if msg.active.is_some() {
		return Err(Error::ProtocolViolation);

Conversation

kixelated commented Sep 26, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Problem

Approach

Impact

Tests

Alternatives

Follow-ups

Uh oh!

chatgpt-codex-connector Bot commented Sep 26, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Codex Review Summary

Uh oh!

chatgpt-codex-connector Bot left a comment

Choose a reason for hiding this comment

💡 Codex Review

Uh oh!

chatgpt-codex-connector Bot Sep 26, 2026

Choose a reason for hiding this comment

Uh oh!

kixelated Sep 27, 2026

Choose a reason for hiding this comment

Uh oh!

kixelated commented Sep 26, 2026

Decisions

Uh oh!

chatgpt-codex-connector Bot left a comment

Choose a reason for hiding this comment

💡 Codex Review

Uh oh!

chatgpt-codex-connector Bot Sep 27, 2026

Choose a reason for hiding this comment

Uh oh!

kixelated Sep 27, 2026

Choose a reason for hiding this comment

Uh oh!

kixelated commented Sep 27, 2026

Merge summary

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

1 participant

kixelated commented Sep 26, 2026 •

edited

Loading

chatgpt-codex-connector Bot commented Sep 26, 2026 •

edited

Loading