quest: settle the moq-mux Clock boundary

[kixelated]

Settles the open decision that review of #4693 added to mux-live-import-clock.

How does std::time::Instant in moq-mux's public API (Clock::at, the json/binary timed writes) handle the switch?

• ✅ Convert at the boundary: public inputs stay std::time::Instant, so it stays on main
• Change the public types: a break on dev

Public API: none. Wire: none.

(Written by Claude Opus 5.5)

🤖 Generated with Claude Code

[kixelated]

MERGE (head 3873e5ddeb6cac2a4e9480de81d32cb384371a87)

Settles the open Instant-boundary choice that #4693 left for the maintainer: convert at the boundary on main (public inputs stay std::time::Instant), reject a typed break on dev. Docs-only; matches the PR decision checklist.

Checks

1. Decision text matches code surface. On the quest base / main, Clock::at(epoch: Instant, wall: SystemTime), capture/stamp, and json/binary update/append take std::time::Instant / Timed<_, Instant>; Clock is re-exported from moq-mux; moq_net::Timed docs still say moq-mux uses std::time::Instant. "Public API: none" is correct for the boundary-convert path.
2. Background facts still hold. ts/fmp4 live_import and flv import still std::thread::sleep(idle) with a 300 ms case; gap goes through Anchor → Clock::now() → std::time::Instant::elapsed(). moq-mux: run the TS SI debounce test on the paused clock #4687 on this line dropped SI Capture's crate::Clock for a private web_async::time::Instant debounce only — Clock itself is still std::time::Instant.
3. No contradictions or broken refs. Settled paragraph is the prior recommended option; rejected alternative is explicit; #4687 / #4693 cites are consistent with the review thread.

Blocking

None.

Non-blocking

1. CI — Quest passed; Check / Test still pending at review time (expected for a single markdown quest settle).
2. Implementer note (inherited, not a defect) — when coding the child, convert in Clock::at and capture/stamp so "ahead of now" / elapsed comparisons stay on one Instant type after the private epoch moves.

This is an automated review, not the maintainer's decision
(Written by Grok)

[kixelated]

Automated review by review (OpenAI)

Reviewed 1ab08a5 (full plan diff; the latest base merge leaves this decision unchanged).

No actionable bugs found. quest/m1/test-flakes-2/mux-live-import-clock.md:33–38 resolves the previous API-boundary prerequisite: preserve the public std::time::Instant inputs while changing the private epoch.

Direction is sound and keeps the change smaller than a caller-wide type migration. The surrounding code confirms both boundaries matter: Clock::at (rs/moq-mux/src/clock.rs:89) and timed JSON/binary writes through Clock::stamp (json.rs:225,309; binary.rs:182,257). The implementation should convert captured instants as well as the constructor epoch, with the future-capture check and elapsed time using the same clock; retain the fixed wall mapping.

Verification: static plan, public API, import-helper, and prior-review inspection only. No builds or tests run. This decision does not yet verify paused-clock behavior or WASM compatibility.

(Written by review (OpenAI))

[kixelated]

Merging. Settles the moq-mux Clock boundary in mux-live-import-clock: public inputs (Clock::at, timed json/binary writes) stay std::time::Instant and convert to the private web_async epoch internally, so no public API change and it stays on main. Merged the line base in; Codex and CI clean. Implementation note from Codex for whoever picks it up: convert captured instants as well as the constructor epoch, so the future-capture check and elapsed time read the same clock.

(Written by Claude Opus 5.5)