From 1a2809d4dec2d08f7e356ab951389117438add3e Mon Sep 17 00:00:00 2001 From: Luke Curley Date: Wed, 14 Oct 2026 10:39:52 -0700 Subject: [PATCH 1/2] quest: settle the untimed follow-ups The maintainer decided #4822's two questions: moq-archive records untimed tracks through an optional timestamp, and a malformed object in a group FETCH ends its track. Each becomes a quest and the decision quest goes. Co-Authored-By: Claude Opus 5.5 --- quest/m1/README.md | 2 +- quest/m1/archive/README.md | 1 + quest/m1/archive/untimed.md | 31 ++++++++++++++++++++++++ quest/m1/group-fetch-malformed.md | 17 ++++++++++++++ quest/m1/untimed-decisions.md | 39 ------------------------------- 5 files changed, 50 insertions(+), 40 deletions(-) create mode 100644 quest/m1/archive/untimed.md create mode 100644 quest/m1/group-fetch-malformed.md delete mode 100644 quest/m1/untimed-decisions.md diff --git a/quest/m1/README.md b/quest/m1/README.md index 87bf0ee398b..ebee59b9dbc 100644 --- a/quest/m1/README.md +++ b/quest/m1/README.md @@ -79,7 +79,7 @@ above them. - [Subscribe-close backport](/quest/m1/subscribe-close-backport.md) - the same fix and the `race` loser fix on `release`, so closing a subscription during setup leaves no unhandled rejection in a later 0.17 patch - [Main live-feed check](/quest/m1/fetch-live-feed.md) - tests pin that a fetch-only or parked warm cache never supplies Largest Object on main, fixing it if it does - [Claim answer epochs](/quest/m1/claim-answer-epochs.md) - a claim answers each output by publishing it at its exact path under its own epoch, so a restarted output never reaches a returning reader as the old one's cache -- [Untimed decisions](/quest/m1/untimed-decisions.md) - the maintainer decides whether moq-archive keeps refusing untimed tracks, and whether a malformed FETCH object ends its track +- [Malformed group FETCH](/quest/m1/group-fetch-malformed.md) - an object missing its Timestamp in a relay's group FETCH ends the whole TIMESCALE track, as subscribe and joining FETCH already do - [FFI shape](/quest/m1/ffi-shape/README.md) - the bindings mirror Rust's layers: net at the root, then media, json, flate, audio, and video namespaces built from the handle below - [Publishing never invents a timestamp](/quest/m1/publish-timestamp.md) - no Rust or binding publish API fills in a timestamp; an untimed payload goes out untimed - [moq-net's async feature](/quest/m1/net-async-feature.md) - async helpers and both sessions sit behind a default `async` feature, and CI builds moq-net without it; Controlled time's net migration requires it diff --git a/quest/m1/archive/README.md b/quest/m1/archive/README.md index 0f4855a54a3..e85effdd511 100644 --- a/quest/m1/archive/README.md +++ b/quest/m1/archive/README.md @@ -167,6 +167,7 @@ owned by that prerequisite, not duplicated in archive storage. - [Replay catalog](/quest/m1/archive/replay-catalog.md) - `moq_archive::Reader` republishes the recorded catalog live with `store` set, so stock `moq export hls` serves the whole replay of `moq import archive` - [Idle flush](/quest/m1/archive/flush.md) - idle tracks are recorded within a bounded wall-clock delay, and `flush()` forces a track's pending record out - [History from the retained floor](/quest/m1/archive/history-floor.md) - history mode lists a retained (DVR) recording from its oldest retained segment instead of refusing it +- [Untimed tracks](/quest/m1/archive/untimed.md) - untimed tracks are recorded and replayed untimed instead of refused, through an optional timestamp in the format ## Related diff --git a/quest/m1/archive/untimed.md b/quest/m1/archive/untimed.md new file mode 100644 index 00000000000..1351e81fe23 --- /dev/null +++ b/quest/m1/archive/untimed.md @@ -0,0 +1,31 @@ +# [M] Untimed tracks are recorded and replayed untimed + +## Goal + +`moq_archive::Writer` records a track with no timescale instead of refusing it +with `Error::Untimed`, so `moq export archive` works over moq-lite before 05 +and moq-transport without TIMESCALE (every track on drafts 14-16). +`moq_archive::Reader` replays such a track untimed: its frames carry no +timestamp, never a send or arrival time standing in for one. A frame missing +its timestamp on a timed track stays malformed. + +## Plan + +Decided 2026-10-14 by the maintainer, from +[#4822](https://github.com/moq-dev/moq/pull/4822)'s untimed follow-ups: bump +the format so a timestamp is optional, rather than keep refusing. + +- Timedness is a track property, so `.info` carries it: an absent `timescale` + means the track's frames store no timestamps. Settle whether that is a new + format version: `moq-archive` is on `release`, so recordings in version 2 + may exist, and the line's in-place exception + ([archive](/quest/m1/archive/README.md), Compatibility) ends once a release + ships recordings. +- The timeline is a time index (`pts`, `duration`, `durationMax`), which an + untimed track can't fill. Decide its shape before coding: likely one record + per group with no `pts` or `duration`, as a sparse track already cuts. +- Consumers that need media time refuse an untimed track loudly: + `moq-hls` can't derive segments from it, and the clock mapping does not + apply. +- Update the draft's Recording section + (`drafts/draft-lcurley-moq-hang.md`) in the same PR. diff --git a/quest/m1/group-fetch-malformed.md b/quest/m1/group-fetch-malformed.md new file mode 100644 index 00000000000..be769088ddf --- /dev/null +++ b/quest/m1/group-fetch-malformed.md @@ -0,0 +1,17 @@ +# [XS] A malformed object in a group FETCH ends its track + +## Goal + +On a TIMESCALE track, a subgroup object without a Timestamp in a standalone +group FETCH, which a relay sends to fill a cache miss (`run_group_fetch` in +`rs/moq-net/src/ietf/subscriber.rs`), ends the whole track with +`MalformedTrack`, as it already does in `recv_group` and a joining FETCH's +fill (`recv_fill`). A regression test pins it. + +## Plan + +Decided 2026-10-14 by the maintainer, from +[#4822](https://github.com/moq-dev/moq/pull/4822)'s untimed follow-ups: the +publisher broke the track however the object arrived, so failing only the +group is not enough. js/net already ends the track +([#4968](https://github.com/moq-dev/moq/pull/4968)). diff --git a/quest/m1/untimed-decisions.md b/quest/m1/untimed-decisions.md deleted file mode 100644 index e55fa11e337..00000000000 --- a/quest/m1/untimed-decisions.md +++ /dev/null @@ -1,39 +0,0 @@ -# [XS] The maintainer decides two untimed follow-ups - -## Goal - -kixelated decides two questions [#4822](https://github.com/moq-dev/moq/pull/4822) -(the untimed model) raised, and each answer becomes a quest or is dropped. -This quest waits on that decision; delete it once both are answered. - -## Plan - -1. **moq-archive and untimed tracks.** `Writer` refuses an untimed track with - `Error::Untimed`, because the format stores a timestamp per frame. So - `moq export archive` over moq-lite before 05, or moq-transport without - TIMESCALE (every track on drafts 14-16), now fails where it used to record - arrival times. - - Recommended: keep refusing. Playback needs media time, and send or - arrival times would mislead a reader. Revisit when someone needs to - archive over those wire versions. - - Alternative: bump the format so a frame's timestamp is optional, and - replay an untimed track untimed. -2. **A malformed object in a FETCH.** On a TIMESCALE track, a subgroup object - without a Timestamp ends the whole track with `MalformedTrack` - (`rs/moq-net/src/ietf/subscriber.rs`, `recv_group`), as js/net does - ([#4968](https://github.com/moq-dev/moq/pull/4968)), and so does one in a - joining FETCH's fill (`recv_fill`). The same object in a standalone group FETCH, which a - relay sends to fill a cache miss (`run_group_fetch`), fails only that group. - - Recommended: end the track too. The publisher broke the track, however - the object arrived. - - Alternative: keep it to the group, so a one-off fetch response can't tear - down a live subscription. - -Settled while planning this (2026-10-07): MOQtail (HEAD `d642d335`) never sends -TIMESCALE as a track property; it puts the scale in its catalog JSON and stamps -objects with LOC's old Timestamp id 0x06. Its tracks arrive untimed, so the -malformed rule can't reach them, and nothing is left to decide there. - -## Related - -- [One max_age meaning](/quest/m1/cache-max-age.md) - the other untimed follow-up, already planned From 4fed6ac3c98472f33a4825d67484d3a50e1bd185 Mon Sep 17 00:00:00 2001 From: Luke Curley Date: Wed, 14 Oct 2026 16:35:35 -0700 Subject: [PATCH 2/2] quest: mirror untimed archive change in js/hang and doc/concept Co-Authored-By: Claude Opus 5.5 --- quest/m1/archive/untimed.md | 6 ++++-- 1 file changed, 4 insertions(+), 2 deletions(-) diff --git a/quest/m1/archive/untimed.md b/quest/m1/archive/untimed.md index 1351e81fe23..1cb31825853 100644 --- a/quest/m1/archive/untimed.md +++ b/quest/m1/archive/untimed.md @@ -27,5 +27,7 @@ the format so a timestamp is optional, rather than keep refusing. - Consumers that need media time refuse an untimed track loudly: `moq-hls` can't derive segments from it, and the clock mapping does not apply. -- Update the draft's Recording section - (`drafts/draft-lcurley-moq-hang.md`) in the same PR. +- Mirror the change in the Hang format and its docs in the same PR: + `js/hang` (`timeline.ts`, the version in `catalog/archive.ts`), the + draft's Recording section (`drafts/draft-lcurley-moq-hang.md`), and + `doc/concept/hang.md`.