Voice 13: fluxer-voice LiveKit room and interop test (#74) #121

Manually merged
charles merged 0 commits from voice-13-voice-livekit-interop into voice 2026-11-12 01:46:26 +00:00
Owner

What

  • crates/fluxer-voice/src/rtc.rs: LiveKitRoom: Room on the LiveKit Rust SDK (livekit 0.10.1) and VoiceSession::livekit(events) -> VoiceHandle.
    • Joins with auto-subscribe off; set_subscribed resolves identity + track sid to the publication and tells the server.
    • Keeps the endpoint's path prefix (signalling_base: the SDK pushes rtc onto the path; a trailing slash, query or fragment is dropped first, so /livekit/ does not become /livekit//rtc).
    • Publishes one microphone track at once, muted, from a native audio source at 48 kHz mono: RED on, DTX off, max_bitrate clamped to 8-510 kbit/s, WebRTC's own processing off.
    • Remote audio from native audio streams at 48 kHz, 10 ms frames (mono; stereo for screen-share audio).
    • Participants, publications, mutes, active speakers, data packets, E2EE state, reconnects and disconnect reasons mapped to RoomEvents.
  • fluxer_voice::apm (Apm, ApmConfig, ApmError): the SDK does expose WebRTC's audio processing module, so it is wrapped with a pure i16 interface (10 ms, 48 kHz, 1 or 2 channels; other shapes are an error, never a panic). No follow-up issue for spec §12 Q3 was needed.
  • crates/fluxer-voice/tests/livekit_interop.rs: two_sessions_hear_a_tone, mute_and_unsubscribe_stop_audio, data_round_trip_on_negotiation_topic, connects_through_path_prefix, plus room_receives_only_subscribed_tracks (below the session, which gates audio by itself: the room must not receive an unsubscribed track). An in-test reverse proxy serves the server under /livekit only and answers 404 elsewhere; tokens are minted with the dev key (livekit-api, dev-dependency, access-token feature only).
  • tools/livekit-test-server.sh: downloads livekit-server v1.12.0 (SHA-256 checked) and runs it in dev mode on loopback.
  • xtask: new last step livekit-interop when FLUXER_LIVEKIT_TESTS=1 (starts the script unless a server is already listening, runs the tests single-threaded, stops it). The cargo test --workspace steps no longer pass that variable on, because -- --ignored (screenshot step) also selects the interop tests, which then skip themselves.
  • CI (Linux job): FLUXER_LIVEKIT_TESTS=1, a "Fetch livekit-server" step, libglib2.0-dev, and clang 21 from apt.llvm.org with CXX=clang++-21.
  • Workspace Cargo.toml: livekit (feature rustls-tls-native-roots), livekit-api. deny.toml needed no change (cargo deny check is green as it is).

Stacked on: voice-12-voice-session, voice-2-proto-api-soundboard-rtc

How it was tested

  • FLUXER_SCREENSHOT_TESTS=1 cargo xtask ci green locally (base 98726e0 was green before any change).
  • 13 new unit tests in rtc.rs that need no server: path prefix, publish options and bitrate clamp, stereo mix-down, token never in errors, reason/source mapping, a room that is not joined, refused E2EE rooms, Apm on real libwebrtc.
  • The interop tests (6) and tools/livekit-test-server.sh have been run since this description was first written: FLUXER_SCREENSHOT_TESTS=1 FLUXER_LIVEKIT_TESTS=1 cargo xtask ci is green locally at 10e1104, with xtask starting livekit-server v1.12.0 through the script. (The first version of this text said they had never been run; that held for commit 6af1f49 only.)
  • Review rounds: 2a9d563 keeps data packets that arrive before their sender is announced (EarlyData, see the module docs); 10e1104 makes sure packets whose sender the SDK named are never dropped with them, and moves the round-trip probe to its own task.

Deviations from the plan

  • Round-trip time: the SDK measures the signalling RTT but does not expose it (livekit-signaling has a TODO for it). The room reports the nominated ICE candidate pair's RTT instead, measured every second; the session still reports every 2 s.
  • E2EE: a Connect with an e2ee_key is refused with an error instead of joining unencrypted; Voice 23 wires the key.
  • Build requirement the plan did not foresee: on Linux, webrtc-sys refuses GCC and needs clang >= 21 (the libc++ inside the prebuilt libwebrtc says so), hence the apt.llvm.org step in CI; Ubuntu 22.04's own clang is 14. Voice 25 (packaging) inherits this.
  • ParticipantJoined is sent on the SDK's ParticipantConnected, which (per the SDK's docs) fires before that participant can receive data messages. The protocol covers it (every client answers the first advertisement of an identity), but an advertisement sent on ParticipantJoined alone may miss a newcomer.
  • fluxer-proto is a dev-dependency of fluxer-voice (the interop test uses CODEC_NEGOTIATION_TOPIC and a real SelectProtocol).
  • The SDK logs through the log crate; nothing bridges log to tracing today, so none of it reaches the log file. If Voice 14 adds a bridge, the SDK's records need the same redaction as ours (Review Focus 7).

Windows / sizes

Not built on Windows here; the non-interop tests have no platform-specific code, but the Windows runner is the check. On Linux the prebuilt libwebrtc is 692 MB unpacked inside the target directory (debug/build/scratch-*/out/livekit_webrtc); the test binaries link it statically and need only libc/libm/libgcc_s at run time. The Windows target-dir size after a clean build still has to be read off the runner: tracked in #127.

Closes #74

🤖 Generated with Claude Code

## What - `crates/fluxer-voice/src/rtc.rs`: `LiveKitRoom: Room` on the LiveKit Rust SDK (`livekit` 0.10.1) and `VoiceSession::livekit(events) -> VoiceHandle`. - Joins with auto-subscribe off; `set_subscribed` resolves identity + track sid to the publication and tells the server. - Keeps the endpoint's path prefix (`signalling_base`: the SDK pushes `rtc` onto the path; a trailing slash, query or fragment is dropped first, so `/livekit/` does not become `/livekit//rtc`). - Publishes one microphone track at once, muted, from a native audio source at 48 kHz mono: RED on, DTX off, `max_bitrate` clamped to 8-510 kbit/s, WebRTC's own processing off. - Remote audio from native audio streams at 48 kHz, 10 ms frames (mono; stereo for screen-share audio). - Participants, publications, mutes, active speakers, data packets, E2EE state, reconnects and disconnect reasons mapped to `RoomEvent`s. - `fluxer_voice::apm` (`Apm`, `ApmConfig`, `ApmError`): the SDK does expose WebRTC's audio processing module, so it is wrapped with a pure `i16` interface (10 ms, 48 kHz, 1 or 2 channels; other shapes are an error, never a panic). No follow-up issue for spec §12 Q3 was needed. - `crates/fluxer-voice/tests/livekit_interop.rs`: `two_sessions_hear_a_tone`, `mute_and_unsubscribe_stop_audio`, `data_round_trip_on_negotiation_topic`, `connects_through_path_prefix`, plus `room_receives_only_subscribed_tracks` (below the session, which gates audio by itself: the room must not receive an unsubscribed track). An in-test reverse proxy serves the server under `/livekit` only and answers 404 elsewhere; tokens are minted with the dev key (`livekit-api`, dev-dependency, `access-token` feature only). - `tools/livekit-test-server.sh`: downloads `livekit-server` v1.12.0 (SHA-256 checked) and runs it in dev mode on loopback. - `xtask`: new last step `livekit-interop` when `FLUXER_LIVEKIT_TESTS=1` (starts the script unless a server is already listening, runs the tests single-threaded, stops it). The `cargo test --workspace` steps no longer pass that variable on, because `-- --ignored` (screenshot step) also selects the interop tests, which then skip themselves. - CI (Linux job): `FLUXER_LIVEKIT_TESTS=1`, a "Fetch livekit-server" step, `libglib2.0-dev`, and clang 21 from apt.llvm.org with `CXX=clang++-21`. - Workspace `Cargo.toml`: `livekit` (feature `rustls-tls-native-roots`), `livekit-api`. `deny.toml` needed no change (`cargo deny check` is green as it is). Stacked on: voice-12-voice-session, voice-2-proto-api-soundboard-rtc ## How it was tested - `FLUXER_SCREENSHOT_TESTS=1 cargo xtask ci` green locally (base `98726e0` was green before any change). - 13 new unit tests in `rtc.rs` that need no server: path prefix, publish options and bitrate clamp, stereo mix-down, token never in errors, reason/source mapping, a room that is not joined, refused E2EE rooms, `Apm` on real libwebrtc. - The interop tests (6) and `tools/livekit-test-server.sh` have been run since this description was first written: `FLUXER_SCREENSHOT_TESTS=1 FLUXER_LIVEKIT_TESTS=1 cargo xtask ci` is green locally at `10e1104`, with xtask starting livekit-server v1.12.0 through the script. (The first version of this text said they had never been run; that held for commit `6af1f49` only.) - Review rounds: `2a9d563` keeps data packets that arrive before their sender is announced (`EarlyData`, see the module docs); `10e1104` makes sure packets whose sender the SDK named are never dropped with them, and moves the round-trip probe to its own task. ## Deviations from the plan - **Round-trip time**: the SDK measures the signalling RTT but does not expose it (`livekit-signaling` has a TODO for it). The room reports the nominated ICE candidate pair's RTT instead, measured every second; the session still reports every 2 s. - **E2EE**: a `Connect` with an `e2ee_key` is refused with an error instead of joining unencrypted; Voice 23 wires the key. - **Build requirement the plan did not foresee**: on Linux, `webrtc-sys` refuses GCC and needs clang >= 21 (the libc++ inside the prebuilt libwebrtc says so), hence the apt.llvm.org step in CI; Ubuntu 22.04's own clang is 14. Voice 25 (packaging) inherits this. - `ParticipantJoined` is sent on the SDK's `ParticipantConnected`, which (per the SDK's docs) fires before that participant can receive data messages. The protocol covers it (every client answers the first advertisement of an identity), but an advertisement sent on `ParticipantJoined` alone may miss a newcomer. - `fluxer-proto` is a dev-dependency of `fluxer-voice` (the interop test uses `CODEC_NEGOTIATION_TOPIC` and a real `SelectProtocol`). - The SDK logs through the `log` crate; nothing bridges `log` to `tracing` today, so none of it reaches the log file. If Voice 14 adds a bridge, the SDK's records need the same redaction as ours (Review Focus 7). ## Windows / sizes Not built on Windows here; the non-interop tests have no platform-specific code, but the Windows runner is the check. On Linux the prebuilt libwebrtc is 692 MB unpacked inside the target directory (`debug/build/scratch-*/out/livekit_webrtc`); the test binaries link it statically and need only libc/libm/libgcc_s at run time. The Windows target-dir size after a clean build still has to be read off the runner: tracked in #127. Closes #74 🤖 Generated with [Claude Code](https://claude.com/claude-code)
feat(voice): session state machine over a room trait
All checks were successful
ci / Linux (x86_64) (push) Successful in 8m1s
ci / Windows (x64) (push) Successful in 8m43s
3622ad7160
New crate fluxer-voice: trait Room, VoiceSession/VoiceHandle, and the
scripted FakeRoom behind the feature "fake".

Refs #73

Co-Authored-By: Claude Fable 5.2 <noreply@anthropic.com>
feat(voice): LiveKit room and interop test
Some checks failed
ci / Linux (x86_64) (push) Failing after 50s
ci / Windows (x64) (push) Has been cancelled
6af1f4984b
LiveKitRoom implements Room on the LiveKit Rust SDK: auto-subscribe off, the
endpoint's path prefix kept, one muted microphone track from a native 48 kHz
source (RED on, DTX off, bitrate clamped to 8-510 kbit/s), remote audio from
native audio streams, room events mapped to RoomEvents, round-trip time from
the nominated ICE candidate pair. VoiceSession::livekit starts a session on it.
fluxer_voice::apm wraps WebRTC's audio processing module behind an i16
interface.

The interop tests run against a local livekit-server v1.12.0 behind an in-test
/livekit reverse proxy when FLUXER_LIVEKIT_TESTS=1; cargo xtask ci starts the
server with tools/livekit-test-server.sh. The Linux CI job installs clang 21
and the glib headers the SDK's build needs.

Refs #74

Co-Authored-By: Claude Fable 5.2 <noreply@anthropic.com>
Author
Owner

Independent review, Voice 13 (commit 6af1f49, diff from 98726e0)

Verdict: changes requested. Two blocking findings; both leave the Linux CI job red. Everything else checked out.

Blocking

  1. data_round_trip_on_negotiation_topic fails deterministically against livekit-server v1.12.0 (10 of 10 runs here). crates/fluxer-voice/tests/livekit_interop.rs:246: user_1001_quiet-otter: the expected event did not come within 3s. The other four interop tests pass every time.
    • Cause, by experiment: a data packet published right after the sender's own VoiceEvent::Connected never arrives. Sleeping 3 s before Bob's PublishData makes the test pass (2 of 2). Sleeping 3 s between Alice's join and Bob's join does not (0 of 2). So the sender is not ready, not the receiver, and publish_data reports no error.
    • This is Review Focus 2, not only a test problem: an advertisement sent on Connected is dropped silently. Either LiveKitRoom::connect must return only once the room can deliver data, or publish_data must wait or retry until it can.
  2. tools/livekit-test-server.sh is committed with mode 100644. The new CI step "Fetch livekit-server" runs it directly (tools/livekit-test-server.sh --fetch), which ends in permission denied. tools/fj.sh is 100755. cargo xtask ci is not affected because it starts the script through bash. Fix: git update-index --chmod=+x tools/livekit-test-server.sh, or call it with bash in the workflow.

Minor

  • Spec §3.1 asks the crate README to record the WebRTC build the SDK fetched. The README gives only the size ("about 700 MB").
  • room_receives_only_subscribed_tracks cannot tell "the server stopped sending" from "the pump stopped reading": the reader task is aborted on TrackUnsubscribed, so zero Audio events follow either way. publication.set_subscribed(false) is called, so the behaviour is probably right; the test just does not prove it (Review Focus 1).
  • Review Focus 2's "again for each newly joined participant" is not exercised at this layer. The PR already says an advertisement sent on ParticipantJoined may miss a newcomer; finding 1 shows the same loss is real on the newcomer's own side.
  • The server script and the SDK's signalling log print the dev-key JWTs at debug level. Harmless here, but it supports the PR's note for Voice 14 (Review Focus 7).

Verified

  • Tests named by the plan: two_sessions_hear_a_tone, mute_and_unsubscribe_stop_audio, data_round_trip_on_negotiation_topic, connects_through_path_prefix all exist, are #[ignore]d and gated on FLUXER_LIVEKIT_TESTS=1. They assert the 440 Hz tone by energy share, mute and unsubscribe stopping it, the round trip on CODEC_NEGOTIATION_TOPIC within 3000 ms, and requests under /livekit/rtc through a proxy that answers 404 elsewhere.
  • Interfaces: LiveKitRoom: Room, VoiceSession::livekit(events) -> VoiceHandle, fluxer_voice::apm (Apm, ApmConfig, ApmError, plain i16). Auto-subscribe off, 48 kHz native source, RED on, DTX off, bitrate clamped to 8 000–510 000 bit/s. No LiveKit type in a public signature.
  • Review Focus 4: livekit-signaling 0.1.9 pushes rtc (and v1) onto the path segments, so the prefix survives; signalling_base strips the trailing slash first. Confirmed live by connects_through_path_prefix.
  • Invariants: fluxer-core untouched (core-isolation step green); livekit-api is a dev-dependency only; the only credentials are the public devkey/secret pair; server-driven paths in rtc.rs have no panics (frame shapes are checked, including before the APM call).
  • Server script: the SHA-256 check and the flags --dev --bind 127.0.0.1 --node-ip 127.0.0.1 work with v1.12.0 on Linux x86_64.

cargo xtask ci (Linux, clang 23.1.4 via CXX=clang++, FLUXER_SCREENSHOT_TESTS=1 FLUXER_LIVEKIT_TESTS=1)

Step Result
fmt, clippy, deny, core-isolation, parity, test, screenshot-tests green
livekit-interop red: 4 passed, 1 failed (finding 1)

Not checked: the Windows build, and the apt.llvm.org clang 21 step on Ubuntu 22.04. Target dir after this run: 22 GB (debug, whole workspace, with both test passes).

## Independent review, Voice 13 (commit 6af1f49, diff from 98726e0) **Verdict: changes requested.** Two blocking findings; both leave the Linux CI job red. Everything else checked out. ### Blocking 1. **`data_round_trip_on_negotiation_topic` fails deterministically against livekit-server v1.12.0** (10 of 10 runs here). `crates/fluxer-voice/tests/livekit_interop.rs:246`: `user_1001_quiet-otter: the expected event did not come within 3s`. The other four interop tests pass every time. - Cause, by experiment: a data packet published right after the sender's own `VoiceEvent::Connected` never arrives. Sleeping 3 s before Bob's `PublishData` makes the test pass (2 of 2). Sleeping 3 s between Alice's join and Bob's join does not (0 of 2). So the sender is not ready, not the receiver, and `publish_data` reports no error. - This is Review Focus 2, not only a test problem: an advertisement sent on `Connected` is dropped silently. Either `LiveKitRoom::connect` must return only once the room can deliver data, or `publish_data` must wait or retry until it can. 2. **`tools/livekit-test-server.sh` is committed with mode 100644.** The new CI step "Fetch livekit-server" runs it directly (`tools/livekit-test-server.sh --fetch`), which ends in `permission denied`. `tools/fj.sh` is 100755. `cargo xtask ci` is not affected because it starts the script through `bash`. Fix: `git update-index --chmod=+x tools/livekit-test-server.sh`, or call it with `bash` in the workflow. ### Minor - Spec §3.1 asks the crate README to record the WebRTC build the SDK fetched. The README gives only the size ("about 700 MB"). - `room_receives_only_subscribed_tracks` cannot tell "the server stopped sending" from "the pump stopped reading": the reader task is aborted on `TrackUnsubscribed`, so zero `Audio` events follow either way. `publication.set_subscribed(false)` is called, so the behaviour is probably right; the test just does not prove it (Review Focus 1). - Review Focus 2's "again for each newly joined participant" is not exercised at this layer. The PR already says an advertisement sent on `ParticipantJoined` may miss a newcomer; finding 1 shows the same loss is real on the newcomer's own side. - The server script and the SDK's signalling log print the dev-key JWTs at debug level. Harmless here, but it supports the PR's note for Voice 14 (Review Focus 7). ### Verified - **Tests named by the plan**: `two_sessions_hear_a_tone`, `mute_and_unsubscribe_stop_audio`, `data_round_trip_on_negotiation_topic`, `connects_through_path_prefix` all exist, are `#[ignore]`d and gated on `FLUXER_LIVEKIT_TESTS=1`. They assert the 440 Hz tone by energy share, mute and unsubscribe stopping it, the round trip on `CODEC_NEGOTIATION_TOPIC` within 3000 ms, and requests under `/livekit/rtc` through a proxy that answers 404 elsewhere. - **Interfaces**: `LiveKitRoom: Room`, `VoiceSession::livekit(events) -> VoiceHandle`, `fluxer_voice::apm` (`Apm`, `ApmConfig`, `ApmError`, plain `i16`). Auto-subscribe off, 48 kHz native source, RED on, DTX off, bitrate clamped to 8 000–510 000 bit/s. No LiveKit type in a public signature. - **Review Focus 4**: `livekit-signaling` 0.1.9 pushes `rtc` (and `v1`) onto the path segments, so the prefix survives; `signalling_base` strips the trailing slash first. Confirmed live by `connects_through_path_prefix`. - **Invariants**: `fluxer-core` untouched (core-isolation step green); `livekit-api` is a dev-dependency only; the only credentials are the public `devkey`/`secret` pair; server-driven paths in `rtc.rs` have no panics (frame shapes are checked, including before the APM call). - **Server script**: the SHA-256 check and the flags `--dev --bind 127.0.0.1 --node-ip 127.0.0.1` work with v1.12.0 on Linux x86_64. ### `cargo xtask ci` (Linux, clang 23.1.4 via `CXX=clang++`, `FLUXER_SCREENSHOT_TESTS=1 FLUXER_LIVEKIT_TESTS=1`) | Step | Result | |---|---| | fmt, clippy, deny, core-isolation, parity, test, screenshot-tests | green | | livekit-interop | **red**: 4 passed, 1 failed (finding 1) | Not checked: the Windows build, and the apt.llvm.org clang 21 step on Ubuntu 22.04. Target dir after this run: 22 GB (debug, whole workspace, with both test passes).
fix(voice): keep data that arrives before its sender is announced
Some checks failed
ci / Windows (x64) (push) Has been cancelled
ci / Linux (x86_64) (push) Has been cancelled
2a9d5634c8
The advertisement a participant sends the moment it is connected reached the
others' data channels before its announcement reached their signalling
connections. The SDK reports such a packet without a sender, and the room
dropped it: data_round_trip_on_negotiation_topic failed against
livekit-server v1.12.0 every time. The room now keeps such packets (2 s, 32
packets, 256 KiB at most) and delivers them as the participant announced
next, 150 ms after the announcement and in order with what that participant
sends meanwhile; a second announcement within the 150 ms drops them, because
their sender is then uncertain.

Also:
- tools/livekit-test-server.sh is executable (the CI step runs it directly).
- room_receives_only_subscribed_tracks proves with the connection's received
  bytes (LiveKitRoom::bytes_received) that the server stops sending an
  unsubscribed track.
- New interop test: data sent on ParticipantJoined reaches the newcomer.
- xtask no longer passes the dev server's debug log (tokens included) on to
  the CI log.
- README names the WebRTC build the SDK fetches (webrtc-a545de4).

Refs #74

Co-Authored-By: Claude Fable 5.2 <noreply@anthropic.com>
Author
Owner

Review round 1: what changed (commit 2a9d563)

FLUXER_SCREENSHOT_TESTS=1 FLUXER_LIVEKIT_TESTS=1 cargo xtask ci is green locally, with the server started by xtask through the script. The interop suite (now 6 tests) passed 10 of 10 further runs against livekit-server v1.12.0.

Blocking

  1. data_round_trip_on_negotiation_topic: fixed in rtc.rs, on the receiving side.
    • The cause is not the one the finding names. The packet is sent and it arrives. It reaches Alice's data channel before Bob's announcement reaches her signalling connection; the SDK then emits DataReceived { participant: None }, and the pump only forwarded packets with a sender. A probe sending one packet every 100 ms from Bob's Connected showed it: packet 0 arrived at Alice without a sender, about 45 ms before her ParticipantConnected(bob), and packets 1 onward arrived normally. That is why a sleep before the send helped and a gap between the joins did not.
    • So neither remedy proposed in the finding (connect returning later, publish_data waiting or retrying) can work: the sender cannot see when the others have learned of it. The server announces Bob when his track is published, which is the moment his connect returns. The original client has the same race and drops the packet (if (topic !== PROTOCOL_TOPIC || !participant) return).
    • The fix: EarlyData in rtc.rs. A packet without a sender is kept (at most 2 s, 32 packets, 256 KiB) and delivered as the participant announced next, 150 ms after that announcement and in order with what that participant sends meanwhile. If a second participant is announced within the 150 ms, the sender is uncertain and the kept packets are dropped, as before.
    • Known limit, documented in the code and the README: the SDK gives neither identity nor sid for such a packet, so the attribution is an inference. A packet from someone announced more than 150 ms after another newcomer, or from the server's API, would count as that newcomer's. The protocol's answer to a first advertisement corrects it. If that is not acceptable, the alternative is to drop as the original does and rely on the answer round; say so and I will change the test instead.
    • Tests: six unit tests of EarlyData (early_data_*), and the interop test unchanged apart from a comment.
  2. tools/livekit-test-server.sh is now 100755. tools/livekit-test-server.sh --fetch run directly works.

Minor

  • README, WebRTC build: recorded. webrtc-sys-build 0.3.21 pins WEBRTC_TAG = webrtc-a545de4; the archive is webrtc-linux-x64-release.zip from the livekit/rust-sdks release of that tag. The archive carries no WebRTC milestone number, so the tag is what is recorded.
  • room_receives_only_subscribed_tracks: fixed. New diagnostic LiveKitRoom::bytes_received() (transport statistics). The test measures what the connection receives in one second while subscribed and again after unsubscribing: 17,484 bytes/s against 34 bytes/s here. It asserts more than 3,000 bytes/s while subscribed and less than a quarter of the subscribed rate afterwards.
  • "Again for each newly joined participant": partly. New interop test data_sent_on_participant_joined_reaches_the_newcomer: Alice sends the moment she gets ParticipantJoined(bob) and Bob receives it within the 3 s grace period (10 of 10). Mapping ParticipantJoined to the SDK's ParticipantActive is not done, because it changes the order of events the session relies on: #125.
  • Tokens in debug logs: the server half is fixed. xtask now discards the server's stderr as well as its stdout; before, about 2,700 lines of its debug log, dev-key tokens included, went into the CI log. The SDK half belongs to Voice 14: #126.
## Review round 1: what changed (commit 2a9d563) `FLUXER_SCREENSHOT_TESTS=1 FLUXER_LIVEKIT_TESTS=1 cargo xtask ci` is green locally, with the server started by xtask through the script. The interop suite (now 6 tests) passed 10 of 10 further runs against livekit-server v1.12.0. ### Blocking 1. **`data_round_trip_on_negotiation_topic`: fixed in `rtc.rs`, on the receiving side.** - The cause is not the one the finding names. The packet is sent and it arrives. It reaches Alice's data channel before Bob's announcement reaches her signalling connection; the SDK then emits `DataReceived { participant: None }`, and the pump only forwarded packets with a sender. A probe sending one packet every 100 ms from Bob's `Connected` showed it: packet 0 arrived at Alice without a sender, about 45 ms before her `ParticipantConnected(bob)`, and packets 1 onward arrived normally. That is why a sleep before the send helped and a gap between the joins did not. - So neither remedy proposed in the finding (`connect` returning later, `publish_data` waiting or retrying) can work: the sender cannot see when the others have learned of it. The server announces Bob when his track is published, which is the moment his `connect` returns. The original client has the same race and drops the packet (`if (topic !== PROTOCOL_TOPIC || !participant) return`). - The fix: `EarlyData` in `rtc.rs`. A packet without a sender is kept (at most 2 s, 32 packets, 256 KiB) and delivered as the participant announced next, 150 ms after that announcement and in order with what that participant sends meanwhile. If a second participant is announced within the 150 ms, the sender is uncertain and the kept packets are dropped, as before. - Known limit, documented in the code and the README: the SDK gives neither identity nor sid for such a packet, so the attribution is an inference. A packet from someone announced more than 150 ms after another newcomer, or from the server's API, would count as that newcomer's. The protocol's answer to a first advertisement corrects it. If that is not acceptable, the alternative is to drop as the original does and rely on the answer round; say so and I will change the test instead. - Tests: six unit tests of `EarlyData` (`early_data_*`), and the interop test unchanged apart from a comment. 2. **`tools/livekit-test-server.sh` is now 100755.** `tools/livekit-test-server.sh --fetch` run directly works. ### Minor - **README, WebRTC build**: recorded. `webrtc-sys-build` 0.3.21 pins `WEBRTC_TAG = webrtc-a545de4`; the archive is `webrtc-linux-x64-release.zip` from the `livekit/rust-sdks` release of that tag. The archive carries no WebRTC milestone number, so the tag is what is recorded. - **`room_receives_only_subscribed_tracks`**: fixed. New diagnostic `LiveKitRoom::bytes_received()` (transport statistics). The test measures what the connection receives in one second while subscribed and again after unsubscribing: 17,484 bytes/s against 34 bytes/s here. It asserts more than 3,000 bytes/s while subscribed and less than a quarter of the subscribed rate afterwards. - **"Again for each newly joined participant"**: partly. New interop test `data_sent_on_participant_joined_reaches_the_newcomer`: Alice sends the moment she gets `ParticipantJoined(bob)` and Bob receives it within the 3 s grace period (10 of 10). Mapping `ParticipantJoined` to the SDK's `ParticipantActive` is not done, because it changes the order of events the session relies on: #125. - **Tokens in debug logs**: the server half is fixed. xtask now discards the server's stderr as well as its stdout; before, about 2,700 lines of its debug log, dev-key tokens included, went into the CI log. The SDK half belongs to Voice 14: #126.
Author
Owner

Independent review, Voice 13, round 2 (commit 2a9d563, diff from 98726e0)

Verdict: changes requested. One blocking finding, in the new EarlyData. Local CI is green and both round-1 blockers are fixed.

Blocking

  1. EarlyData discards packets whose sender is known (crates/fluxer-voice/src/rtc.rs, EarlyData::announced, with EarlyData::named).
    • named() queues a newcomer's properly attributed packets in the same Claim::packets as the kept unnamed ones. A second announcement within the 150 ms then does self.claim.take() and throws the whole claim away, named packets included.
    • Proven with a temporary unit test (not committed): unnamed(1), announced("user_1_a"), named("user_1_a", 2) returns None, announced("user_2_b") 50 ms later, and settle never hands out packet 2.
    • Why it matters (Review Focus 2): when several people join at once, Bob's answer to Alice's advertisement and his advertisement for Carol both reach Alice with a sender, inside Bob's 150 ms window. If Carol is announced in that window, Alice loses all of them and never learns Bob's codecs. Before this commit those packets were delivered.
    • Fix: when the sender becomes uncertain, drop only the unnamed packets and deliver the named ones of the claim (in order). Add a unit test for it.

Decision asked of the reviewer: attribution by inference

Accepted, once finding 1 is fixed. I confirmed in livekit 0.10.1 (Room::handle_data) that the SDK resolves the sender against known participants and dispatches participant: None otherwise, so no identity is available. The bounds (2 s, 32 packets, 256 KiB, 150 ms settle) are reasonable, the only topic this client interprets is codec negotiation, and a wrong attribution is overwritten by the sender's next advertisement. Dropping as the original does would fail Review Focus 2 for the newcomer's first advertisement.

Minor

  • The PR description still says the interop tests were not run and the script is unexecuted. Update it; comment 3537 is the only place that says otherwise.
  • Pump::run awaits measure_rtt (up to 1 s, every 1 s) inside the select arm. If get_stats stalls, for example while reconnecting, room events and data wait behind it.
  • The plan asks for the Windows target-dir size in the PR; still open, needs the runner.
  • ParticipantJoined on ParticipantConnected (#125) and the SDK's token-bearing log records (#126) are deferred with linked follow-ups; both exist and are labelled.

Verified

  • Plan tests: two_sessions_hear_a_tone (energy share at 440 Hz, nothing before subscribing), mute_and_unsubscribe_stop_audio, data_round_trip_on_negotiation_topic (real SelectProtocol on CODEC_NEGOTIATION_TOPIC, within 3000 ms, both directions), connects_through_path_prefix (proxy answers 404 outside /livekit; requests seen under /livekit/rtc). All #[ignore]d and gated on FLUXER_LIVEKIT_TESTS=1. Plus room_receives_only_subscribed_tracks (now with the byte-rate check) and data_sent_on_participant_joined_reaches_the_newcomer.
  • Interfaces: LiveKitRoom: Room, VoiceSession::livekit(events) -> VoiceHandle, fluxer_voice::apm (Apm, ApmConfig, ApmError, plain i16). Auto-subscribe off, native source at 48 kHz, RED on, DTX off, bitrate clamped to 8 000-510 000 bit/s. No LiveKit type in a public signature.
  • Invariants: fluxer-core untouched; livekit-api is a dev-dependency only; the only credentials are the public dev pair; server-driven paths in rtc.rs do not panic; EarlyData byte accounting cannot underflow.
  • Round 1: tools/livekit-test-server.sh is 100755; the README records webrtc-a545de4; the CI log of my run contains no token.

cargo xtask ci (Linux, clang 23.1.4)

Run Result
cargo xtask ci green (screenshot and interop steps skipped)
FLUXER_SCREENSHOT_TESTS=1 FLUXER_LIVEKIT_TESTS=1 cargo xtask ci green; livekit-interop 6 passed in 19.8 s against livekit-server v1.12.0 started by xtask

Not checked: the Windows build and the remote runners.

## Independent review, Voice 13, round 2 (commit 2a9d563, diff from 98726e0) **Verdict: changes requested.** One blocking finding, in the new `EarlyData`. Local CI is green and both round-1 blockers are fixed. ### Blocking 1. **`EarlyData` discards packets whose sender is known** (`crates/fluxer-voice/src/rtc.rs`, `EarlyData::announced`, with `EarlyData::named`). - `named()` queues a newcomer's properly attributed packets in the same `Claim::packets` as the kept unnamed ones. A second announcement within the 150 ms then does `self.claim.take()` and throws the whole claim away, named packets included. - Proven with a temporary unit test (not committed): `unnamed(1)`, `announced("user_1_a")`, `named("user_1_a", 2)` returns `None`, `announced("user_2_b")` 50 ms later, and `settle` never hands out packet 2. - Why it matters (Review Focus 2): when several people join at once, Bob's answer to Alice's advertisement and his advertisement for Carol both reach Alice with a sender, inside Bob's 150 ms window. If Carol is announced in that window, Alice loses all of them and never learns Bob's codecs. Before this commit those packets were delivered. - Fix: when the sender becomes uncertain, drop only the unnamed packets and deliver the named ones of the claim (in order). Add a unit test for it. ### Decision asked of the reviewer: attribution by inference Accepted, once finding 1 is fixed. I confirmed in `livekit` 0.10.1 (`Room::handle_data`) that the SDK resolves the sender against known participants and dispatches `participant: None` otherwise, so no identity is available. The bounds (2 s, 32 packets, 256 KiB, 150 ms settle) are reasonable, the only topic this client interprets is codec negotiation, and a wrong attribution is overwritten by the sender's next advertisement. Dropping as the original does would fail Review Focus 2 for the newcomer's first advertisement. ### Minor - The PR description still says the interop tests were not run and the script is unexecuted. Update it; comment 3537 is the only place that says otherwise. - `Pump::run` awaits `measure_rtt` (up to 1 s, every 1 s) inside the select arm. If `get_stats` stalls, for example while reconnecting, room events and data wait behind it. - The plan asks for the Windows target-dir size in the PR; still open, needs the runner. - `ParticipantJoined` on `ParticipantConnected` (#125) and the SDK's token-bearing log records (#126) are deferred with linked follow-ups; both exist and are labelled. ### Verified - **Plan tests**: `two_sessions_hear_a_tone` (energy share at 440 Hz, nothing before subscribing), `mute_and_unsubscribe_stop_audio`, `data_round_trip_on_negotiation_topic` (real `SelectProtocol` on `CODEC_NEGOTIATION_TOPIC`, within 3000 ms, both directions), `connects_through_path_prefix` (proxy answers 404 outside `/livekit`; requests seen under `/livekit/rtc`). All `#[ignore]`d and gated on `FLUXER_LIVEKIT_TESTS=1`. Plus `room_receives_only_subscribed_tracks` (now with the byte-rate check) and `data_sent_on_participant_joined_reaches_the_newcomer`. - **Interfaces**: `LiveKitRoom: Room`, `VoiceSession::livekit(events) -> VoiceHandle`, `fluxer_voice::apm` (`Apm`, `ApmConfig`, `ApmError`, plain `i16`). Auto-subscribe off, native source at 48 kHz, RED on, DTX off, bitrate clamped to 8 000-510 000 bit/s. No LiveKit type in a public signature. - **Invariants**: `fluxer-core` untouched; `livekit-api` is a dev-dependency only; the only credentials are the public dev pair; server-driven paths in `rtc.rs` do not panic; `EarlyData` byte accounting cannot underflow. - **Round 1**: `tools/livekit-test-server.sh` is 100755; the README records `webrtc-a545de4`; the CI log of my run contains no token. ### `cargo xtask ci` (Linux, clang 23.1.4) | Run | Result | |---|---| | `cargo xtask ci` | green (screenshot and interop steps skipped) | | `FLUXER_SCREENSHOT_TESTS=1 FLUXER_LIVEKIT_TESTS=1 cargo xtask ci` | green; livekit-interop 6 passed in 19.8 s against livekit-server v1.12.0 started by xtask | Not checked: the Windows build and the remote runners.
fix(voice): never drop early data whose sender is known
Some checks failed
ci / Windows (x64) (push) Failing after 7m7s
ci / Linux (x86_64) (push) Has been cancelled
10e11040c7
A second arrival within the settle window dropped the whole claim,
including the packets the first newcomer had sent under its own name
while its unnamed ones waited. Only the packets without a sender are in
doubt: the named ones are now delivered at once, in order, when the
claim is given up (second arrival, or the participant leaving).

The round-trip probe moves to a task of its own, so that statistics
that stall (while reconnecting) no longer hold up room events and data.

Refs #74

Co-Authored-By: Claude Fable 5.2 <noreply@anthropic.com>
Author
Owner

Review round 2: what changed (commit 10e1104)

FLUXER_SCREENSHOT_TESTS=1 FLUXER_LIVEKIT_TESTS=1 cargo xtask ci is green locally (interop suite included, server started by xtask through the script).

Blocking

  1. EarlyData discards packets whose sender is known: fixed.
    • Claim now keeps the two kinds apart: kept (the packets without a sender, taken for the newcomer's) and sent (what the newcomer sent under its own name while they waited).
    • A second announcement inside the settle window drops only kept (and the unnamed packets that arrived since, as before). EarlyData::announced returns the first newcomer's sent packets and the pump delivers them at once, in order, before the second ParticipantJoined. Nothing waits after that, so the first newcomer's later packets pass straight through behind them.
    • Same hole, same remedy, when the claimed participant leaves inside the window: EarlyData::left returns its sent packets and they are delivered before ParticipantLeft (they were dropped too).
    • settle is unchanged for the reader: kept then sent.
    • Tests: new packets_with_a_known_sender_survive_a_second_arrival (the finding's sequence: unnamed(1), announced(a), named(a, 2), named(a, 3), unnamed(4), announced(b) at +50 ms yields (a, [2, 3]), nothing later, and both pass straight through afterwards); early_data_of_a_participant_that_left_is_dropped extended for the leaving case; early_data_is_dropped_when_two_participants_arrive_at_once now asserts that nothing is handed out when the first one sent nothing.

Minor

  • Attribution by inference: no change, as accepted. The doc comment of EarlyData now states that packets with a named sender are never dropped.
  • Pump::run awaited measure_rtt inside the select arm: fixed. The probe is a task of its own (probe_rtt, in the pump's JoinSet, so it is aborted with the pump) that sends RoomEvent::Rtt itself; statistics that stall no longer hold up room events or data packets. Interval and timeout are unchanged.
  • Stale PR description: "How it was tested" is being corrected to say that the interop tests and the server script are run (see this comment and 3537 until it is). The Windows target-dir size still needs the runner: follow-up #127.
  • ParticipantConnected vs ParticipantActive (#125), SDK log records with the token (#126): no change, deferred to those follow-ups.
## Review round 2: what changed (commit 10e1104) `FLUXER_SCREENSHOT_TESTS=1 FLUXER_LIVEKIT_TESTS=1 cargo xtask ci` is green locally (interop suite included, server started by xtask through the script). ### Blocking 1. **`EarlyData` discards packets whose sender is known: fixed.** - `Claim` now keeps the two kinds apart: `kept` (the packets without a sender, taken for the newcomer's) and `sent` (what the newcomer sent under its own name while they waited). - A second announcement inside the settle window drops only `kept` (and the unnamed packets that arrived since, as before). `EarlyData::announced` returns the first newcomer's `sent` packets and the pump delivers them at once, in order, before the second `ParticipantJoined`. Nothing waits after that, so the first newcomer's later packets pass straight through behind them. - Same hole, same remedy, when the claimed participant leaves inside the window: `EarlyData::left` returns its `sent` packets and they are delivered before `ParticipantLeft` (they were dropped too). - `settle` is unchanged for the reader: `kept` then `sent`. - Tests: new `packets_with_a_known_sender_survive_a_second_arrival` (the finding's sequence: `unnamed(1)`, `announced(a)`, `named(a, 2)`, `named(a, 3)`, `unnamed(4)`, `announced(b)` at +50 ms yields `(a, [2, 3])`, nothing later, and both pass straight through afterwards); `early_data_of_a_participant_that_left_is_dropped` extended for the leaving case; `early_data_is_dropped_when_two_participants_arrive_at_once` now asserts that nothing is handed out when the first one sent nothing. ### Minor - **Attribution by inference**: no change, as accepted. The doc comment of `EarlyData` now states that packets with a named sender are never dropped. - **`Pump::run` awaited `measure_rtt` inside the select arm: fixed.** The probe is a task of its own (`probe_rtt`, in the pump's `JoinSet`, so it is aborted with the pump) that sends `RoomEvent::Rtt` itself; statistics that stall no longer hold up room events or data packets. Interval and timeout are unchanged. - **Stale PR description**: "How it was tested" is being corrected to say that the interop tests and the server script are run (see this comment and 3537 until it is). The Windows target-dir size still needs the runner: follow-up #127. - **`ParticipantConnected` vs `ParticipantActive` (#125), SDK log records with the token (#126)**: no change, deferred to those follow-ups.
Author
Owner

Independent review, Voice 13, round 3 (commit 10e1104, diff from 98726e0)

Verdict: approved. No blocking findings. Local CI is green, interop step included. The Windows build and the remote runners are not checked here.

Blocking

None.

Minor

  1. Windows is still unverified. Nothing in this PR has been built on Windows, and the plan's "target-dir size after a clean build" is not in the PR. Tracked in #127 (open, label follow-up); the Windows runner is the only check that webrtc-sys builds there.
  2. The round-trip probe has no test of its own. probe_rtt (now a task in the pump's JoinSet) is covered only by the interop suite running with it. By reading: it is aborted with the pump, ends when the session stops listening, and no longer sits in the select arm.
  3. RoomEvent::Rtt is documented in room.rs as the signalling connection's round trip, while LiveKitRoom reports the nominated ICE pair's. The deviation is stated in the PR, the README and rtc.rs; the enum's doc comment is the one place that still says otherwise.
  4. #120 ("interop tests have not been run") is obsolete but still open without a comment. The tests run in cargo xtask ci and pass. For Charles to close.
  5. Deferred with linked follow-ups, unchanged: ParticipantJoined on ParticipantConnected (#125), the SDK's token-bearing log records (#126).

Round 2 finding: fixed

  • Claim separates kept (packets without a sender) from sent (packets the SDK attributed to the newcomer). A second announcement inside the 150 ms drops only kept and the unnamed queue; EarlyData::announced returns sent, and the pump delivers it before the second ParticipantJoined.
  • The same applies when the claimed participant leaves inside the window: EarlyData::left returns sent, delivered before ParticipantLeft.
  • packets_with_a_known_sender_survive_a_second_arrival runs the sequence from the finding and asserts (user_1_a, [2, 3]), nothing handed out later, and straight pass-through afterwards. It passes.
  • Byte accounting still cannot underflow (bytes is reset wherever the queue is cleared or drained).

Verified

  • Plan tests: two_sessions_hear_a_tone (energy share at 440 Hz in 100 ms windows; no audio before subscribing), mute_and_unsubscribe_stop_audio, data_round_trip_on_negotiation_topic (real SelectProtocol on CODEC_NEGOTIATION_TOPIC, both directions, each within 3000 ms), connects_through_path_prefix (proxy answers 404 outside /livekit; request seen at /livekit/rtc; trailing slash does not produce //). All #[ignore]d and gated on FLUXER_LIVEKIT_TESTS=1. Extra: room_receives_only_subscribed_tracks, data_sent_on_participant_joined_reaches_the_newcomer.
  • Interfaces: LiveKitRoom: Room, VoiceSession::livekit(events) -> VoiceHandle, fluxer_voice::apm (Apm, ApmConfig, ApmError, plain i16). Auto-subscribe off; native source at 48 kHz; RED on, DTX off; max_bitrate from RoomOptions, clamped to 8 000-510 000 bit/s; remote audio from native streams at 48 kHz; active speakers, track and participant events, data, RTT mapped. No LiveKit type in a public signature.
  • Review Focus 4: signalling_base keeps the prefix and strips trailing slash, query and fragment; proven live by connects_through_path_prefix. Review Focus 2 (this task's half): advertisement on connect and on ParticipantJoined both arrive within 3000 ms.
  • Invariants: fluxer-core untouched (core-isolation green); fluxer-audio untouched; livekit-api is a dev-dependency with access-token only; the only credentials are the public devkey/secret pair; server-driven paths in rtc.rs do not panic (frame shapes, track sids, RTT values and APM shapes are checked); the SDK error text has the token struck out; deny.toml unchanged and cargo deny green.
  • Housekeeping: script is 100755; all three commits carry Refs #74 and the attribution line; the PR description now says the interop tests are run.

FLUXER_SCREENSHOT_TESTS=1 FLUXER_LIVEKIT_TESTS=1 cargo xtask ci (Linux, clean target dir)

Step Result
fmt, clippy, deny, core-isolation, parity, test, screenshot-tests green
livekit-interop green: 6 passed in 19.7 s against livekit-server v1.12.0 started by xtask

The CI log of this run contains no token.

## Independent review, Voice 13, round 3 (commit 10e1104, diff from 98726e0) **Verdict: approved.** No blocking findings. Local CI is green, interop step included. The Windows build and the remote runners are not checked here. ### Blocking None. ### Minor 1. **Windows is still unverified.** Nothing in this PR has been built on Windows, and the plan's "target-dir size after a clean build" is not in the PR. Tracked in #127 (open, label `follow-up`); the Windows runner is the only check that `webrtc-sys` builds there. 2. **The round-trip probe has no test of its own.** `probe_rtt` (now a task in the pump's `JoinSet`) is covered only by the interop suite running with it. By reading: it is aborted with the pump, ends when the session stops listening, and no longer sits in the select arm. 3. **`RoomEvent::Rtt` is documented in `room.rs` as the signalling connection's round trip**, while `LiveKitRoom` reports the nominated ICE pair's. The deviation is stated in the PR, the README and `rtc.rs`; the enum's doc comment is the one place that still says otherwise. 4. **#120 ("interop tests have not been run") is obsolete but still open without a comment.** The tests run in `cargo xtask ci` and pass. For Charles to close. 5. Deferred with linked follow-ups, unchanged: `ParticipantJoined` on `ParticipantConnected` (#125), the SDK's token-bearing log records (#126). ### Round 2 finding: fixed - `Claim` separates `kept` (packets without a sender) from `sent` (packets the SDK attributed to the newcomer). A second announcement inside the 150 ms drops only `kept` and the unnamed queue; `EarlyData::announced` returns `sent`, and the pump delivers it before the second `ParticipantJoined`. - The same applies when the claimed participant leaves inside the window: `EarlyData::left` returns `sent`, delivered before `ParticipantLeft`. - `packets_with_a_known_sender_survive_a_second_arrival` runs the sequence from the finding and asserts `(user_1_a, [2, 3])`, nothing handed out later, and straight pass-through afterwards. It passes. - Byte accounting still cannot underflow (`bytes` is reset wherever the queue is cleared or drained). ### Verified - **Plan tests**: `two_sessions_hear_a_tone` (energy share at 440 Hz in 100 ms windows; no audio before subscribing), `mute_and_unsubscribe_stop_audio`, `data_round_trip_on_negotiation_topic` (real `SelectProtocol` on `CODEC_NEGOTIATION_TOPIC`, both directions, each within 3000 ms), `connects_through_path_prefix` (proxy answers 404 outside `/livekit`; request seen at `/livekit/rtc`; trailing slash does not produce `//`). All `#[ignore]`d and gated on `FLUXER_LIVEKIT_TESTS=1`. Extra: `room_receives_only_subscribed_tracks`, `data_sent_on_participant_joined_reaches_the_newcomer`. - **Interfaces**: `LiveKitRoom: Room`, `VoiceSession::livekit(events) -> VoiceHandle`, `fluxer_voice::apm` (`Apm`, `ApmConfig`, `ApmError`, plain `i16`). Auto-subscribe off; native source at 48 kHz; RED on, DTX off; `max_bitrate` from `RoomOptions`, clamped to 8 000-510 000 bit/s; remote audio from native streams at 48 kHz; active speakers, track and participant events, data, RTT mapped. No LiveKit type in a public signature. - **Review Focus 4**: `signalling_base` keeps the prefix and strips trailing slash, query and fragment; proven live by `connects_through_path_prefix`. **Review Focus 2** (this task's half): advertisement on connect and on `ParticipantJoined` both arrive within 3000 ms. - **Invariants**: `fluxer-core` untouched (core-isolation green); `fluxer-audio` untouched; `livekit-api` is a dev-dependency with `access-token` only; the only credentials are the public `devkey`/`secret` pair; server-driven paths in `rtc.rs` do not panic (frame shapes, track sids, RTT values and APM shapes are checked); the SDK error text has the token struck out; `deny.toml` unchanged and `cargo deny` green. - **Housekeeping**: script is 100755; all three commits carry `Refs #74` and the attribution line; the PR description now says the interop tests are run. ### `FLUXER_SCREENSHOT_TESTS=1 FLUXER_LIVEKIT_TESTS=1 cargo xtask ci` (Linux, clean target dir) | Step | Result | |---|---| | fmt, clippy, deny, core-isolation, parity, test, screenshot-tests | green | | livekit-interop | green: 6 passed in 19.7 s against livekit-server v1.12.0 started by xtask | The CI log of this run contains no token.
build(windows): link the static C runtime, as libwebrtc does
Some checks failed
ci / Linux (x86_64) (push) Has been cancelled
ci / Windows (x64) (push) Has been cancelled
7c28fba480
The Windows link of fluxer-voice failed with LNK2038 (MT_StaticRelease against
MD_DynamicRelease) and LNK2005 between libcpmt and msvcprt.

Refs #74, #128

Co-Authored-By: Claude Fable 5.2 <noreply@anthropic.com>
Merge remote-tracking branch 'origin/voice' into voice-13-voice-livekit-interop
Some checks failed
ci / Windows (x64) (push) Failing after 16m53s
ci / Linux (x86_64) (push) Successful in 29m32s
1678f3eb69
Merge remote-tracking branch 'origin/voice' into voice-13-voice-livekit-interop
All checks were successful
ci / Windows (x64) (push) Successful in 25m45s
ci / Linux (x86_64) (push) Successful in 31m19s
555028702d
charles manually merged commit 53e3d60b8b into voice 2026-11-12 01:46:26 +00:00
Sign in to join this conversation.
No description provided.