Tags: livekit/client-sdk-unity
Tags
Final README updates for 2.0.0 (#349) * Removing developer preview warning * Adding a screenshot of meet * other image path * Adding youtube tutorial screenshot, adding samples links * Adding tutorial section * More meet readme * Adding agents sample readme * Add description text * Removing duplicate samples mention * Less white space, put async to bottom, new snippet for connect * Aligned with docs * Improved data and byte stream examples * Agent targeting IL2CPP and ARM64 * Rpc method done * Publishing camera track * Split up audio section * Updating to TokenSource changes * Adding error log to file not found * Add real youtube tutorial link * docs: use read-first incremental stream reader pattern The ByteStreamReader/TextStreamReader incremental examples used a Reset-first loop (Reset → yield → break on IsEos → read), which breaks out at end-of-stream before draining chunks that are still buffered. Switch the documented coroutine examples to the read-first pattern (yield → break on IsEos → read → Reset), which drains every buffered chunk before stopping. Accompanies the reader fix in the following commit. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> --------- Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Update ffi 0.12.68 and prepare for Unity v1.4.0 (#334) * Run update script * Show usage of DegradationPreference * Rename FramePacketTrailer to FrameMetadata * Stashing dynacast false * Go up to 1.4.0 We have shuffled around some files and this is breaking
Return a PublishDataInstruction from PublishData (#301) * Add PlayMode tests for PublishData size limits LocalParticipant.PublishData is fire-and-forget — the C# API has no callback, so server-side drops of oversized packets are silent. These tests probe the boundary empirically by publishing payloads of varying sizes between two participants and asserting delivery via the subscriber's DataReceived event. Retries on a 200 ms interval to avoid SFU warm-up flakiness. Verified against livekit-server 1.12.0: 1 KiB and 15 KiB arrive intact (length + bytes); 65 KiB is dropped. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> * Return a PublishDataInstruction from PublishData PublishData was fire-and-forget (void), so callers could not tell whether the Rust side accepted the packet. The FFI transport already carried the result end-to-end — PublishDataRequest has a request_async_id, Rust emits a PublishDataCallback { async_id, error }, and the data_task forwards the publish result into it — but the C# side discarded the event. Wire it through: - New PublishDataInstruction (YieldInstruction) exposing IsError + Error. - The three PublishData overloads now return it instead of void (source-compatible: callers ignoring the result still compile). - Route the callback by adding PublishData to FFIClient.ExtractRequestAsyncId. Packets over the negotiated maximum message size now surface the error ("data packet size (N bytes) exceeds the negotiated maximum message size (64000 bytes)") via instruction.Error. Tests: ReturnsSuccess_ForSmallPayload verifies the success path on any binary. ReturnsError_ForOversizedPayload (Explicit) verifies the size-limit error and requires the client-sdk-rust datamessage_size FFI binary. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * Clean up old simple tests * No ignoring messages since it is only warn --------- Co-authored-by: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Max/prepare 1.3.7 release (#283) * Adding logic to listen to release or rust-dependency label * Executed prepare release script * Prepare release script also bumps submodule recursively * Changes to DataTrack proto included * Renamed bundle identifier * Check for null object on teardown * Use default token config
PreviousNext