Skip to content
Merged
Show file tree
Hide file tree
Changes from all commits
Commits
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
12 changes: 6 additions & 6 deletions quest/README.md
Original file line number Diff line number Diff line change
Expand Up @@ -7,12 +7,12 @@ grouped into milestones ordered by priority.

## Plan

m0 is everything in flight now: relay hardening and IETF interop ahead of
Seattle, wildcard routing, broadcast epochs, and audio playout (jitter target
m0 is everything in flight now: relay hardening and the last known IETF
interop fix ahead of Seattle, wildcard routing, broadcast epochs, and audio playout (jitter target
and quality harness). m1 is the next wave across reliability, features, performance, and
planning. m2 holds later features, design studies, and experiments. m3 is
gated on the outside world: hardware, a partner, a consumer, or a provider's
offer. m4 waits on an upstream release.
gated on the outside world (hardware, a partner, a consumer, or a provider's
offer) or is speculative work with no named consumer yet. m4 waits on an upstream release.

A quest waiting on the outside world, in any milestone, requires a small quest
beside it that names the condition. That condition quest stays ready, so
Expand All @@ -21,8 +21,8 @@ move the blocked quest to the milestone its priority belongs in.

## Required

- [m0: immediate priorities](/quest/m0/README.md) - everything in flight now: relay hardening and IETF interop for Seattle, wildcard routing, broadcast epochs, and audio playout
- [m0: immediate priorities](/quest/m0/README.md) - everything in flight now: relay hardening and the last known IETF interop fix for Seattle, wildcard routing, broadcast epochs, and audio playout
- [m1: next wave](/quest/m1/README.md) - reliability, capabilities, performance, and the planning that settles their contracts
- [m2: later work](/quest/m2/README.md) - deferred features, design studies, and experiments
- [m3: deferred](/quest/m3/README.md) - gated on the outside world: hardware, a partner, a consumer, or a provider's offer
- [m3: deferred](/quest/m3/README.md) - gated on the outside world (hardware, a partner, a consumer, or a provider's offer), or speculative with no named consumer
- [m4: upstream](/quest/m4/README.md) - waiting on an upstream release, re-checked periodically
12 changes: 8 additions & 4 deletions quest/m0/README.md
Original file line number Diff line number Diff line change
Expand Up @@ -20,10 +20,12 @@ done. moq.pro pins this repository's `release` line, so the release gate below
also keeps #4741 from reaching it early.
The Pronto GPU integration lives in moq.pro.

The branch flip ranks first: it is a short cutover that every later PR
targets. Relay hardening: IETF interop leads the rest, since only those
quests block Seattle. IETF stream types came from m1 in the 2026-09-30 audit
because a session ended by legal input is exactly what Seattle would hit. The DoS hardening
The branch flip is done except its #4605 backport to `release`; the
Cloudflare switch waits on the maintainer as a condition quest. Relay
hardening: [REQUEST_OK accepts LARGEST_OBJECT](/quest/m0/ietf-largest-object.md)
leads the rest, since it is the one known case left of legal moq-transport
input closing a session, which Seattle would hit. The 2026-10-05 audit split
it from the m2 validator findings, which block nothing. The DoS hardening
from an external review on 2026-09-29, verified against `main`, stays in m0
as security work. Its quests describe fixes, not exploits.

Expand All @@ -48,6 +50,8 @@ a published `@moq/watch` break.
## Required

- [Branch flip](/quest/m0/branch-flip.md) - `dev` becomes the default `main` trunk and today's `main` becomes `release`, where publishing runs
- [Cloudflare builds track release](/quest/m0/cloudflare-release.md) - condition: the maintainer points the docs and demo builds at `release`
- [REQUEST_OK accepts LARGEST_OBJECT](/quest/m0/ietf-largest-object.md) - a conformant moq-transport peer's REQUEST_OK no longer closes the session, ahead of Seattle
- [Request caps](/quest/m0/request-caps.md) - lite message sizes, IETF request IDs, and per-session announces and subscriptions are bounded
- [noq reassembly cap](/quest/m0/noq-reassembly-cap.md) - iroh's upstream noq carries quinn's stream reassembly cap, once n0 releases it
- [qmux credit](/quest/m0/qmux-credit.md) - qmux returns connection credit for dropped and stopped streams and delivers its close frame, on both lines
Expand Down
21 changes: 13 additions & 8 deletions quest/m0/branch-flip.md
Original file line number Diff line number Diff line change
@@ -1,4 +1,4 @@
# [S] Trunk is main, releases ship from release
# [XS] Trunk is main, releases ship from release

## Goal

Expand Down Expand Up @@ -89,14 +89,19 @@ its ruleset has no linear history, the `release` ruleset exists, and the `dev`
ruleset is gone. The first back-merge ran during the rename window and 404ed;
its re-run opened #4742.

Done (verified in the 2026-10-05 audit): #4742 landed on `main` as a merge
commit, release-plz #4596 ran on `release`, and the back-merges #4762, #4797,
and #4831 landed on `main`.

Remaining:

- The maintainer points the Cloudflare docs (`moq-doc`) and demo builds'
production branch at `release`.
- Once #4605 merges on `main`, cherry-pick it into `release` as a backport PR.
- Verify: #4742 lands on `main` as a merge commit; release-plz runs on
`release`; a chained publish builds `release` while trunk differs; the
back-merges of two patch publishes on `release`, without a cut between,
both land on `main`.
- #4605 merged on `main` (275af9352, 2026-10-03), but `release` still lacks
its 64 MiB default receive window. Cherry-pick it into `release` as a
backport PR.

Public API: none. Wire: none. Contributors see the new branch model.

## Related

- [Cloudflare builds track release](/quest/m0/cloudflare-release.md) - the maintainer's dashboard step, waiting on its own condition
- [noq reassembly cap](/quest/m0/noq-reassembly-cap.md) - the rest of the cap on `release`
18 changes: 12 additions & 6 deletions quest/m0/broadcast-epoch/README.md
Original file line number Diff line number Diff line change
Expand Up @@ -52,12 +52,17 @@ Decided:
prefix route. Document this rather than promise it works.
- Publishers on the default publish path, such as moq-boy and moq-room,
inherit the epoch from Origin. moq-stats mints its own through
[Stats epochs](/quest/m1/stats-epoch.md), which also gates the release
[Stats epochs](/quest/m0/broadcast-epoch/stats-epoch.md), which also gates the release
(decided 2026-10-04): a restarted stats node under a reused name stalls its
viewers the same way.
- [Retracted demand release](/quest/m1/unannounce-demand-release.md) also
- [Retracted demand release](/quest/m0/broadcast-epoch/unannounce-demand-release.md) also
gates the release (decided 2026-10-05): a regression from #4741 on main that
`release` lacks.
- Decided in the 2026-10-05 audit: the m1 quests gating this line (stats
epochs, the bounded stats aggregate it requires, and retracted demand
release) moved under it, and the OBS half of GStreamer and OBS moved to m1
as [OBS publishes under epochs](/quest/m1/obs-epoch.md), so the release
gate no longer waits on m1 work.
- Derived output mirrors the epoch it came from
(`.pro/transcode/<pid>/foo.hang/@e`, per the
[wildcard](/quest/m0/wildcard/README.md) line's derived-output layout), so
Expand All @@ -75,12 +80,13 @@ This README owns:

## Required

- [Retracted demand release](/quest/m0/broadcast-epoch/unannounce-demand-release.md) - a retracted broadcast's track demand is released when its last subscriber leaves, as before #4741
- [Origin](/quest/m0/broadcast-epoch/origin.md) - moq-net publish mints an epoch, consumers follow the newest live one, and bare requests resolve to it on every version
- [Apps](/quest/m0/broadcast-epoch/apps.md) - moq-cli, the browser publish and watch components, and demo/web publish under epochs and play bare names
- [Gateways](/quest/m0/broadcast-epoch/gateways.md) - RTMP, SRT, and WHIP ingest mint an epoch per incoming connection, so an encoder reconnect is a clean takeover
- [TS restart](/quest/m0/broadcast-epoch/ts-restart.md) - a signalled backward TS discontinuity finishes the broadcast and continues the same input under a fresh epoch
- [Bindings](/quest/m0/broadcast-epoch/bindings.md) - moq-ffi and every wrapper expose the epoch and inherit the default
- [GStreamer and OBS](/quest/m0/broadcast-epoch/gst-obs.md) - moqsink and the OBS plugin publish each run under a fresh epoch
- [Remove `--hop`](/quest/m0/broadcast-epoch/hop-removal.md) - `moq` takes an optional `--epoch` instead of `--hop`, a plain publisher declares a random Hop ID, and the per-session hop stamp and NO_CAPACITY are gone
- [Stats epochs](/quest/m1/stats-epoch.md) - moq-stats publishes each node under its own epoch, so a restarted node never stalls its viewers
- [Retracted demand release](/quest/m1/unannounce-demand-release.md) - a retracted broadcast's track demand is released when its last subscriber leaves, as before #4741
- [GStreamer](/quest/m0/broadcast-epoch/gst.md) - moqsink publishes each run under a fresh epoch
- [Remove `--hop`](/quest/m0/broadcast-epoch/hop-removal.md) - `moq` takes an optional `--epoch` instead of `--hop`, a plain publisher declares a random Hop ID, and the per-session hop stamp is gone
- [Bounded stats aggregate](/quest/m0/broadcast-epoch/stats-aggregate-bound.md) - the stats aggregator folds departed nodes into a retired total, so epoch churn stops growing its memory
- [Stats epochs](/quest/m0/broadcast-epoch/stats-epoch.md) - moq-stats publishes each node under its own epoch, so a restarted node never stalls its viewers
5 changes: 5 additions & 0 deletions quest/m0/broadcast-epoch/bindings.md
Original file line number Diff line number Diff line change
Expand Up @@ -19,6 +19,11 @@ reconnect counter `session.epoch()` is renamed so "epoch" has one meaning.

Decided in the 2026-09-30 audit: libmoq is frozen (renamed `rs/moq-c`), so C and C++ consumers get epochs from moq-ffi.

Decided in the 2026-10-05 audit: this lands before the
[FFI shape](/quest/m1/ffi-shape/README.md) line (#4519), which reshapes the
same wrappers and keeps `epoch()` today. It rebases onto this quest and
adopts the epoch surface and the rename, so the wrappers break once each.

## Required

- [Origin](/quest/m0/broadcast-epoch/origin.md) - the behavior the bindings surface
20 changes: 0 additions & 20 deletions quest/m0/broadcast-epoch/gst-obs.md

This file was deleted.

27 changes: 27 additions & 0 deletions quest/m0/broadcast-epoch/gst.md
Original file line number Diff line number Diff line change
@@ -0,0 +1,27 @@
# [S] GStreamer publishes under epochs

## Goal

`moqsink` publishes each run under a fresh epoch, so a restarted pipeline is
a clean takeover for viewers.

## Plan

`moqsink` takes the origin default per session. When
[#3115](/quest/m2/3115-moqsink-the-publication-has-no-generation-so-a-flush.md)
lands, each of its publication generations is a new epoch. Update
`doc/bin/gstreamer.md` if it shows paths.

Decided in the 2026-10-05 audit: the OBS half moved to
[OBS publishes under epochs](/quest/m1/obs-epoch.md), so the release gate no
longer waits on the C++ line. `moqsink` is Rust on moq-net and needs nothing
from it. moq-c announces through `broadcast::Producer::announce`, so it only
inherits an epoch if Origin mints one there; the OBS quest checks that.

## Required

- [Origin](/quest/m0/broadcast-epoch/origin.md) - the publish default moqsink takes

## Related

- [OBS publishes under epochs](/quest/m1/obs-epoch.md) - the OBS plugin's half, after the C++ line
21 changes: 10 additions & 11 deletions quest/m0/broadcast-epoch/hop-removal.md
Original file line number Diff line number Diff line change
Expand Up @@ -9,7 +9,7 @@ default); a redundant pair passes the same value. A plain publisher declares
a random Hop ID per process with no flag, as the bindings and js already do,
so a multi-homed publisher still catches its own loops. `--cluster-id` names
a node and no longer falls back to `--hop`. Relays stop stamping an unnamed
publisher's hop chain, and NO_CAPACITY is gone: every refusal is terminal.
publisher's hop chain.

## Plan

Expand All @@ -21,18 +21,17 @@ publisher's hop chain, and NO_CAPACITY is gone: every refusal is terminal.
`js/net/src/hop.ts` and its callers in `js/net/src/{lite,ietf}/subscriber.ts`,
and their tests, including `rs/moq-net/tests/legacy_reconnect.rs`. Keep the
leading 0, so an unnamed chain still ranks anonymous.
- Remove NO_CAPACITY entirely (decided 2026-10-03: the maintainer does not
want a retryable refusal). Every refusal is terminal, and an advertiser
sheds load by withdrawing or re-pricing its route. That covers
`StreamCode.NoCapacity` and its uses in `js/net/src/{error,origin}.ts` and
the reserved-code comments in `rs/moq-net/src/error.rs`.
- NO_CAPACITY's removal (decided 2026-10-03: every refusal is terminal)
lands with the [wildcard](/quest/m0/wildcard/README.md) line, whose branch
already deletes it from js/net, moq-net, and both drafts (found in the
2026-10-05 audit). Don't redo it here; rebase onto that line if it lands
first.
- Drafts: lite-07 and cluster-02 are in progress, so edit their text and
changelogs in place. In `drafts/draft-lcurley-moq-lite.md`: the stamping
rule, NO_CAPACITY in Error Codes and Resolution, "the identity it stamps",
rule, "the identity it stamps",
and "the first entry identifies the endpoint that originated the route". In
`drafts/draft-lcurley-moq-cluster.md`: the unknown-publisher stamping text,
the relay behavior that stamps, NO_CAPACITY and its IANA row, and the
changelog bullets.
the relay behavior that stamps, and the changelog bullets.
- CLI: `rs/moq-cli/src/{args,complete,fetch,ls}.rs`. Relays keep `cluster.id`
as their node id.
- Docs: `doc/bin/cli.md` ("Redundant publishers" uses `--epoch`), the
Expand All @@ -46,8 +45,8 @@ idle timeout (a failover service that retracts it, or active-active delivery
to the relay). Neither is required to land this.

Public API: removes `--hop`, `MOQ_HOP`, and `--origin`, adds `--epoch`, and
`--cluster-id` stops reading `--hop`; js/net loses `StreamCode.NoCapacity`.
Wire: relays stop stamping, and lite-07 and cluster-02 lose NO_CAPACITY.
`--cluster-id` stops reading `--hop`.
Wire: relays stop stamping.

## Required

Expand Down
8 changes: 5 additions & 3 deletions quest/m0/broadcast-epoch/origin.md
Original file line number Diff line number Diff line change
Expand Up @@ -32,9 +32,11 @@ any wire change.
exclusion and the per-path fronts from #3312 stay intact.
- Benchmark resolution swept over epochs per name and names per origin, so
following does not scan the table.
- Open: does a catalog `broadcast` reference by bare name pin the epoch its
catalog came from, or follow the newest? Settle it with
[Catalog track alias](/quest/m1/catalog-track-alias.md).
- Settle here, in this m0 quest, whether a catalog `broadcast` reference by
bare name pins the epoch its catalog came from or follows the newest
(decided in the 2026-10-05 audit: it is not handed to
[Catalog track alias](/quest/m1/catalog-track-alias.md), an m1 quest that
never mentions epochs). Record the answer in the line README.
- Update `doc/concept` and `drafts/draft-lcurley-moq-lite.md` wherever they
describe resolution or takeover. The rule is a relay behavior, so state it
in the draft even though no field changes.
Expand Down
Original file line number Diff line number Diff line change
Expand Up @@ -9,7 +9,7 @@ Today `aggregate::Merged` keeps a sticky traffic node with `Reader::Ended`
forever (the `STICKY` unannounce branch in `rs/moq-stats/src/aggregate.rs`).
A returning path reuses its entry, but every distinct path ever announced (a
scale-out, a renamed node) stays in a long-lived aggregator's `nodes`. This is
a bug on main regardless of [stats epochs](/quest/m1/stats-epoch.md), which
a bug on main regardless of [stats epochs](/quest/m0/broadcast-epoch/stats-epoch.md), which
make it worse by giving every restart a distinct path.

## Plan
Expand Down Expand Up @@ -40,5 +40,5 @@ Public API: none expected beyond a possible grace-window knob on

## Related

- [Stats epochs](/quest/m1/stats-epoch.md) - every restarted node gets a new name, so retired entries churn faster
- [Stats epochs](/quest/m0/broadcast-epoch/stats-epoch.md) - every restarted node gets a new name, so retired entries churn faster
- [Binary delta stats](/quest/m2/stats-delta.md) - also sweeps the aggregate over node count
Original file line number Diff line number Diff line change
Expand Up @@ -17,7 +17,7 @@ broadcast name or the group numbers a relay cached under it.
- Decide what an unset node becomes once the epoch follows it (#4739 used
`local`).
- The aggregator treats a new epoch as a new node: its counters add to the
retired total from the [bounded aggregate](/quest/m1/stats-aggregate-bound.md)
retired total from the [bounded aggregate](/quest/m0/broadcast-epoch/stats-aggregate-bound.md)
instead of regressing the merged total.
- MoQ Pro wants to share the same epoch with its storage stats, so expose it
from `Producer`.
Expand All @@ -37,4 +37,4 @@ names gain a trailing epoch segment.

## Required

- [Bounded stats aggregate](/quest/m1/stats-aggregate-bound.md) - retired nodes fold into a bounded total, which epochs churn
- [Bounded stats aggregate](/quest/m0/broadcast-epoch/stats-aggregate-bound.md) - retired nodes fold into a bounded total, which epochs churn
9 changes: 9 additions & 0 deletions quest/m0/cloudflare-release.md
Original file line number Diff line number Diff line change
@@ -0,0 +1,9 @@
# [XS] Cloudflare docs and demo builds track release

## Goal

Condition: the maintainer points the Cloudflare docs (`moq-doc`) and demo
builds' production branch at `release` in the Cloudflare dashboard, as the
[branch flip](/quest/m0/branch-flip.md) decided on 2026-10-02. Check: the
dashboard's production branch for both projects reads `release`, and the
live docs match `release`, not `main`. Delete this quest once it does.
30 changes: 30 additions & 0 deletions quest/m0/ietf-largest-object.md
Original file line number Diff line number Diff line change
@@ -0,0 +1,30 @@
# [S] REQUEST_OK accepts LARGEST_OBJECT

## Goal

On moq-transport, a REQUEST_OK or REQUEST_UPDATE_OK carrying LARGEST_OBJECT
(0x09) is accepted in `rs/moq-net` and `js/net`, instead of closing the
session with PROTOCOL_VIOLATION. The draft makes the parameter a MUST when
objects exist, so a conformant peer hits this on its first reply. It lands
before Seattle interop on 2026-10-12.

## Plan

Decided in the 2026-10-05 audit: split from [moq-transport request
codes](/quest/m2/ietf-request-codes.md), whose other cases stay in m2,
because a session closed by legal input is exactly what m0's relay hardening
rules out.

- `RequestOk::decode_msg` (`rs/moq-net/src/ietf/request.rs`) allows only
0x08 and ACTIVE_COUNT today. Decode 0x09 as a `Location`, which
`ietf/parameters.rs` already handles, and decide whether anything reads it
or it is ignored like EXPIRES.
- Mirror in `js/net/src/ietf`.
- A regression test per language that fails on `main`: a REQUEST_OK with
LARGEST_OBJECT decodes, on each draft that carries it.

Public API: none. Wire: none new; a legal reply stops closing the session.

## Related

- [moq-transport request codes](/quest/m2/ietf-request-codes.md) - the rest of the validator's request-level findings
6 changes: 4 additions & 2 deletions quest/m0/noq-reassembly-cap.md
Original file line number Diff line number Diff line change
Expand Up @@ -15,8 +15,10 @@ and released, then bump `noq` / `noq-proto` (directly or through `iroh`) on
## Plan

`moq-noq` already carries the cap (2.0.1 on `main`, 1.3.3 on `release`), and
`moq-tokio` defaults the connection receive window to 64 MiB on noq and iroh,
so only this pin remains.
`moq-tokio` on `main` defaults the connection receive window to 64 MiB on noq
and iroh (#4605). `release` lacks that default until the
[branch flip](/quest/m0/branch-flip.md)'s backport of #4605 lands, so the bump
on `release` goes in after it.

## Related

Expand Down
2 changes: 1 addition & 1 deletion quest/m0/shared-fronts.md
Original file line number Diff line number Diff line change
Expand Up @@ -56,4 +56,4 @@ Public API: none. Wire: none.

- [Front deadlines](/quest/m1/front-deadline-index.md) - per-front cost per track
- [Front parking](/quest/m1/origin-front-parks.md) - also changes what mints a front
- [Wildcard](/quest/m0/wildcard/README.md) - its line rewrites `model/origin.rs` heavily (+401 lines, fronts end on a standing refusal); land after it or rebase onto it
- [Wildcard](/quest/m0/wildcard/README.md) - its line edits `model/origin.rs` resolution, and its [refusal child](/quest/m0/wildcard/refusal-final.md) makes a standing refusal end the front; land after it or rebase onto it
Loading
Loading