Version: 5.x.x

Migration Guide

This guide covers what you need to change when upgrading the Flutter SDK from v4.0.1 to v5.0.0. Many names changed, several calls changed shape, actions now throw errors you must catch, and Web, macOS and Windows are no longer supported.

Update the dependency first:

dependencies:
  videosdk: ^5.0.0

Then run flutter clean and flutter pub get.

caution

v5.0.0 changes where your app can run and how it builds:

Android and iOS only. Web, macOS and Windows are not supported. If your app ships on any of them, stay on v4.0.1.
Android: set minSdkVersion to 24 or higher (v4.0.1 needed 23). The SDK runs on arm64-v8a and armeabi-v7a only, not on x86 and x86_64 emulators. Test on an ARM device or an arm64 emulator image. To keep Google Play from offering your app to those devices, add ndk { abiFilters += listOf("arm64-v8a", "armeabi-v7a") } to defaultConfig in android/app/build.gradle.kts (ndk { abiFilters 'arm64-v8a', 'armeabi-v7a' } in a Groovy build.gradle), and add disable-abi-filtering=true to android/gradle.properties (Flutter 3.41 or later). Without that line, Flutter adds its own ABIs to the filter and yours has no effect.
iOS: the plugin now ships as a Swift Package and needs Flutter 3.41 or later. Turn on Flutter's Swift Package Manager support for the project by adding config: with enable-swift-package-manager: true under flutter: in your app's pubspec.yaml, or run flutter config --enable-swift-package-manager on every machine that builds your app, including CI.

Sections 1 to 7 cover the changes that stop your app compiling, plus a few in the same areas that still compile but behave differently, such as setting track.enabled or reading RTCVideoRenderer.textureId (section 1) or checking a track factory's result for null (section 3). Section 6 lists every renamed API. Read section 3 even if your app compiles: on Android the SDK no longer asks for camera or microphone permission, so your app must ask before it joins. Sections 8 to 12 compile without errors but behave differently at runtime, so the analyzer will not lead you to them. Section 13 lists what was removed.

1. Remove `videosdk_webrtc`

v5.0.0 no longer uses videosdk_webrtc, and your app must not depend on it either: on iOS it clashes with the SDK. RTCVideoView, RTCVideoRenderer, RTCVideoViewObjectFit and MediaDeviceInfo keep their names and now come from package:videosdk/videosdk.dart, so delete the dependency and the import.

The media objects changed with it. A participant's stream is now RTCStream (it was Stream), and CustomTrack is now CustomStreamTrack. The members you reached through videosdk_webrtc are gone.

Before

import 'package:videosdk/videosdk.dart';
import 'package:videosdk_webrtc/flutter_webrtc.dart';

Future<void> switchCamera(CustomTrack cameraTrack) async {
  final MediaStreamTrack videoTrack =
      cameraTrack.mediaStream.getVideoTracks().first;
  await Helper.switchCamera(videoTrack);
}

After

import 'package:videosdk/videosdk.dart';

Future<void> switchCamera(Room room) async {
  await room.changeCam(); // no arguments: the next camera
}

v4.0.1	v5.0.0
`Helper.switchCamera(track)`	`room.changeCam()` in a room. Before joining, stop the track and create a new one with the other `facingMode`.
`track.enabled = false` to mute	`room.muteMic()` or `room.disableCam()`. Setting `enabled` no longer mutes anything.
`mediaStream.dispose()`, `track.stop()`	`customTrack.stop()`
`stream.track.pause()`, `stream.track.resume()`	`stream.pause()`, `stream.resume()`
`stream.track.getStats()`	`participant.getVideoStats()`, `getAudioStats()` or `getShareStats()`
`stream.track.peerId`	`stream.participantId`
`stream.track` and its other members, such as `track`, `stream`, `rtpReceiver`, `rtpParameters`, `spatialLayer` and `close()`	Removed. Use `stream.id`, `stream.kind` and `stream.participantId`.
`RTCVideoRenderer.textureId` in a `Texture` widget	Always `null`. Render with `RTCVideoView`.

Key points:

Remove videosdk_media_effects too, because it depends on videosdk_webrtc. For noise suppression, pass noiseSuppression in createMicrophoneAudioTrack(noiseConfig:) (section 3). A background image has no replacement on Android. On iOS, room.setVideoProcessor() takes the image's file path or URL (section 7). Background blur is new in VideoSDK.setVideoEffects().
MediaStream and MediaStreamTrack can no longer be imported. Where you annotated cameraTrack.mediaStream with one of them, let Dart infer it: final media = cameraTrack.mediaStream;.
v5.0.0 no longer pulls in random_string, sdp_transform, h264_profile_level_id, flutter_foreground_task, http, encrypt, videosdk_room_stats, videosdk_otel, dart_webrtc, device_info_plus, platform_detect, socket_io_client, protobuf or fixnum, nor the packages they brought with them, such as shared_preferences, path_provider, crypto and logging. If your code imports a package that is not in your own pubspec.yaml, add it there: after flutter pub get, dart analyze reports each one as Target of URI doesn't exist.
flutter_foreground_task also added WAKE_LOCK, RECEIVE_BOOT_COMPLETED and POST_NOTIFICATIONS to your app's merged Android manifest. v5.0.0 does not, so declare any of them that your own code uses in android/app/src/main/AndroidManifest.xml.
RTCVideoRenderer.initialize() now completes at once. Keeping the call is harmless.

2. `VideoSDK.createRoom()` parameters changed

Three parameters were removed, three were renamed, and maxResolution changed type and default.

Before

final room = VideoSDK.createRoom(
  roomId: roomId,
  token: token,
  displayName: "Alex",
  customCameraVideoTrack: cameraTrack,
  customMicrophoneAudioTrack: micTrack,
  maxResolution: 'hd',
  defaultCameraIndex: 1, // the front camera in the 4.x examples
  debugMode: true,
);

After

final room = VideoSDK.createRoom(
  roomId: roomId,
  token: token,
  participantName: "Alex",
  customCameraTrack: cameraTrack,
  customMicrophoneTrack: micTrack,
  maxResolution: MaxResolution.HD,
  defaultCamera: FacingMode.FRONT,
);

Parameter	v4.0.1	v5.0.0
`displayName`	`String`, required	Renamed `participantName`, and optional. Leave it out and the SDK picks a random name.
`customCameraVideoTrack`, `customMicrophoneAudioTrack`	`CustomTrack?`	Renamed `customCameraTrack` and `customMicrophoneTrack`, of type `CustomStreamTrack?`
`maxResolution`	`String`, default `'sd'`	`MaxResolution` enum. Left out, the SDK uses `MaxResolution.HD`.
`mode`	`Mode`, including `Mode.CONFERENCE` and `Mode.VIEWER`	`Mode.SEND_AND_RECV`, `Mode.RECV_ONLY` or `Mode.SIGNALLING_ONLY`. `CONFERENCE` and `VIEWER` are removed.
`participantId`, `multiStream`, `mode`, `metaData`, `signalingBaseUrl`, `preferredProtocol`	Dart default values	Optional, with no Dart default. Left out, the SDK applies its own default, the same as before.
`defaultCameraIndex`	`int`, default `0`	Removed. Use `defaultCamera`.
`defaultCamera`	—	`FacingMode`, default `FacingMode.FRONT`
`debugMode`	`bool`, default `true`	Removed, with no replacement. `VideoSDK.setLogLevel()` sets how much the SDK logs.
`notification`, `NotificationInfo`	Text of the Android screen-share notification	Removed. The SDK shows its own notification.

Key points:

setQuality() now gets a higher layer by default. In v4.0.1 the camera always opened at 1280x720, and maxResolution (default 'sd') only chose the layer setQuality() received: "high" got the middle one. The v5.0.0 default, MaxResolution.HD, keeps the camera at 1280x720 and gives "high" the top layer, so remote video uses more bandwidth. MaxResolution.SD keeps the v4.0.1 layers, but also opens the camera at 960x540.
defaultCamera picks the camera by direction, not by its position in the device list. In v4.0.1 the list from VideoSDK.getVideoDevices() starts with a back camera on Android and iOS phones, so the default index 0 usually opened a back camera. v5.0.0 opens the front camera unless you pass defaultCamera: FacingMode.BACK.
To see the SDK's own logs in flutter run, call await VideoSDK.enableLogForwarding((level, message) => debugPrint('[$level] $message'));.

3. Track factories are typed and throw

createCameraVideoTrack() and createMicrophoneAudioTrack() now return Future<CustomStreamTrack> instead of dynamic. CustomTrack is renamed CustomStreamTrack, and its configs are renamed too: CustomVideoTrackConfig is CustomCameraTrackConfig, and CustomAudioTrackConfig is CustomMicrophoneTrackConfig. Two behaviour changes come with that: a failure throws VideoSDKError instead of returning null and reporting on VideoSDK.on(Events.error), and on Android the SDK no longer asks the user for permission.

Before

VideoSDK.on(Events.error, (error) {
  log("Could not create the track: ${error['message']}");
});

// Asked for camera permission itself, and returned null on failure.
CustomTrack cameraTrack = await VideoSDK.createCameraVideoTrack(
  encoderConfig: CustomVideoTrackConfig.h720p_w1280p,
  facingMode: FacingMode.user,
);

After

final permissions = await VideoSDK.requestPermissions(Permissions.VIDEO);
if (permissions['video'] != true) return;

final CustomStreamTrack cameraTrack;
try {
  cameraTrack = await VideoSDK.createCameraVideoTrack(
    encoderConfig: CustomCameraTrackConfig.h720p_w1280p,
    facingMode: FacingMode.FRONT,
  );
} on VideoSDKError catch (error) {
  // For example 3017 (permission denied), 3021 (no camera), 3023 (camera in use).
  log("Could not create the track: ${error.code} ${error.name}");
  return;
}

Key points:

Ask for permission before you join. On Android the SDK never asks for camera or microphone permission: not in the track factories, and not when join(), unmuteMic(), enableCam() or a changeMode() to Mode.SEND_AND_RECV turns a device on. Without permission that device stays off and the SDK reports 3017 (camera) or 3018 (microphone). v4.0.1 asked in all of these places. Call VideoSDK.requestPermissions() before you create tracks, call join(), or call changeMode(Mode.SEND_AND_RECV).
Without permission the factories throw 3017 (camera) or 3018 (microphone). Other codes: 3021 and 3022 for no camera or microphone, 3023 and 3024 for a device in use, 3026 when the camera fails to open, and 3071 for an unknown cameraId or microphoneId.
noiseConfig is a NoiseConfig and decides every flag. Pass NoiseConfig(noiseSuppression: true, echoCancellation: true, autoGainControl: true) instead of a map. In v4.0.1, a flag you left out of createMicrophoneAudioTrack(noiseConfig:) took its value from the encoderConfig profile. In v5.0.0, once you pass a NoiseConfig, every flag you leave out is turned on, except the new highPassFilter, which stays off. Pass all of noiseSuppression, echoCancellation and autoGainControl, or pass no noiseConfig to keep the profile's values.
maxLayer takes EncodingLayer.MAX_LAYER_2 or EncodingLayer.MAX_LAYER_3 instead of an int.
CustomStreamTrack.stop() replaces dispose(). The new onEnded callback tells you when the track ends, with a VideoSDKError when the device was lost or access was revoked.

enableScreenShare() no longer takes enableAudio or source. Create a screen-share track with VideoSDK.createScreenShareVideoTrack() and pass it as customTrack.

Before

await room.enableScreenShare(enableAudio: true);

After

final CustomStreamTrack? shareTrack = await VideoSDK.createScreenShareVideoTrack(
  encoderConfig: CustomScreenShareTrackConfig.h720p_15fps,
  withAudio: true, // Android only
);
// On Android, null means the user declined the capture prompt.
if (shareTrack != null) {
  await room.enableScreenShare(customTrack: shareTrack);
}

Key points:

On Android, createScreenShareVideoTrack() asks for capture consent and returns null if the user declines. On iOS, one enableScreenShare() call opens the system broadcast picker.
withAudio works on Android only. iOS screen share sends no audio.
enableScreenShare() without a track still works. It shares at CustomScreenShareTrackConfig.h720p_15fps and asks for consent itself. To use another profile, create a track.
source and Room.getScreenShareSources() chose a desktop screen or window. Desktop is not supported, and getScreenShareSources() is removed.
iOS screen share still needs the Broadcast Upload Extension described in Flutter iOS Screen Share. Use the extension files that page lists: an older extension built on ScreenShareRing.swift sends no video with v5.0.0.

5. PubSub `subscribe()` takes a listener object

subscribe() now takes the topic, one PubSubMessageListener that carries all four callbacks, and an optional PubSubSubscribeOptions. It returns nothing: stored messages arrive on onOldMessagesReceived. Failures throw VideoSDKError instead of the PubSubException types.

Before

Future<void> openChat() async {
  try {
    await room.pubSub.subscribe(
      "CHAT",
      onMessage,
      options: const PubSubSubscribeOptions(oldMessageLimit: 50),
      onOldMessagesReceived: (batch, isLast) {
        setState(() => messages.insertAll(0, batch));
      },
    );
  } on PubSubException catch (e) {
    log("Subscribe failed: ${e.name}");
  }
}

Future<void> closeChat() => room.pubSub.unsubscribe("CHAT", onMessage);

After

late final chatListener = PubSubMessageListener(
  onMessageReceived: (PubSubMessage message) {
    setState(() => messages.add(message));
  },
  onOldMessagesReceived: (List<PubSubMessage> batch, PubSubHistoryInfo info) {
    setState(() => messages.insertAll(0, batch));
  },
);

Future<void> openChat() async {
  try {
    await room.pubSub.subscribe(
      "CHAT",
      chatListener,
      const PubSubSubscribeOptions(oldMessageLimit: 50),
    );
  } on VideoSDKError catch (e) {
    log("Subscribe failed: ${e.code} ${e.name}");
  }
}

// Pass the same listener object to unsubscribe.
Future<void> closeChat() => room.pubSub.unsubscribe("CHAT", chatListener);

v4.0.1	v5.0.0
`subscribe(topic, onMessage, {options, onOldMessagesReceived, onBatchReceived, onMessageDrop})` returning `PubSubMessages`	`subscribe(topic, PubSubMessageListener(...), [PubSubSubscribeOptions])` returning `Future<void>`
`onOldMessagesReceived: (messages, bool isLast)`	`onOldMessagesReceived: (messages, PubSubHistoryInfo info)`, with `info.isLast`
`onMessageDrop: (int droppedCount)`	`onMessageDrop: (PubSubMessageDropInfo info)`, with `info.droppedCount`
`PubSubRealtimeOverflow`	`RealtimeOverflow.QUEUE` or `RealtimeOverflow.DROP`
`unsubscribe(topic, onMessage)`	`unsubscribe(topic, listener)`, with the same listener object
`PubSubMessages`	Removed
`PubSubMessage.timestamp` (`DateTime`)	A `String`, as the server sends it
`PubSubException` and its five subtypes	`VideoSDKError`
`PubSubMessage.sendOnly` on received messages	Removed. `PubSubPublishOptions(sendOnly: ...)` for sending is unchanged.

Key points:

Each listener is its own subscription, with its own options and its own history. Subscribing the same listener to a topic twice throws VideoSDKError 3073.
A PubSub call made before the room is joined throws a VideoSDKError with code 3035, so subscribe from your Events.roomJoined handler. While the room reconnects, a call can throw VideoSDKError 3065.
RealtimeStore.observe(key, listener) no longer returns an observer id. Stop with stopObserving(key, listener), passing the same function. RealtimeStore calls made before join also throw a VideoSDKError with code 3035. In v4.0.1 they reported 3022 on Events.error and returned null or an empty string.

6. Renamed APIs

Many APIs have new names. The old names are removed, not deprecated, so the analyzer points you to each one.

Before

room.on(Events.participantLeft, (String participantId, LeaveReason reason) {
  if (reason == LeaveReason.removePeer) showRemovedBanner(participantId);
});
await room.changeCam(cameras.first);
await participant.setQuality("high");

After

room.on(Events.participantLeft, (Participant participant, LeaveReason reason) {
  if (reason == LeaveReason.REMOVE_PEER) showRemovedBanner(participant.id);
});
await room.changeCam(deviceId: cameras.first.deviceId);
await participant.setQuality(VideoQuality.HIGH);

`VideoSDK`

v4.0.1	v5.0.0
`VideoSDK.getVideoDevices()`, `getAudioDevices()`	`VideoSDK.getCameras()`, `getMics()`. `getDevices()` returns both.
`VideoDeviceInfo`, `AudioDeviceInfo`	`CameraDeviceInfo`, `MicrophoneDeviceInfo`
`room.selectedCam`, `selectedMic`, `selectedSpeaker` (and the `*Id` getters)	`VideoSDK.getSelectedVideoDevice()`, `VideoSDK.getSelectedAudioDevice()`. They report the devices in use from join on, and are `null` before join.
`VideoSDK.logLevel`	`VideoSDK.getLogLevel()`
`VideoSDK.keyProvider`	`VideoSDK.getKeyProvider()`
`BaseKeyProvider.setKey(participantId, key)`	`VideoSDK.setParticipantKey(participantId, key)`. A `null` key removes it, and so does the new `BaseKeyProvider.removeKey(participantId)` on the installed provider.
`BaseKeyProvider(discardFrameWhenCryptorNotReady: ...)`	`BaseKeyProvider()`. Remove the argument; nothing replaces it.
`VideoSDK.applyVideoProcessor(videoProcessorName:)`, `VideoSDK.removeVideoProcessor()`	`room.setVideoProcessor(name)`, `room.removeVideoProcessor()`. Both work before `join()`.

`Room`

v4.0.1	v5.0.0
`micEnabled`, `camEnabled`	`isMicEnabled`, `isCamEnabled`
`activeSpeakerId`, `activePresenterId`	`getActiveSpeakerId()`, `getActivePresenterId()`
`livestreamState`	`liveStreamState`
`hlsDownstreamUrl`, `hlsUrls`	`hlsUrl`, an `HLSUrl` with `playbackHlsUrl` and `livestreamUrl`
`changeCam(device)`	`changeCam(deviceId: device.deviceId)`, or `changeCam(customCameraTrack: track)`
`changeMic(device)`	`changeMic(deviceId: device.deviceId)`, or `changeMic(customMicrophoneTrack: track)`
`enableCam(customVideoTrack)`, `unmuteMic(customAudioTrack)`	The optional track is a `CustomStreamTrack`.
`switchAudioDevice(device)`	`setAudioDevice(device)`
`startHls()`, `stopHls()`	`startHLS()`, `stopHLS()`
`config` maps on `startRecording()`, `startHls()`, `startLivestream()`	`RecordingConfig`, `HLSConfig`, `LivestreamConfig`, with `ConfigLayout`
`startLivestream(outputs)` with `{'url', 'streamKey'}` maps	`startLivestream([LivestreamOutput(url: ..., streamKey: ...)])`
`postTranscriptionConfig:` on `startRecording()`, `startHls()`	`transcription:`
`startLivestream(outputs, postTranscriptionConfig:)`	`startLivestream(outputs, config:)`. The transcription argument is removed.
`startTranscription(transcriptionConfig:)`	`startTranscription(config:)`
`requestMediaRelay(meetingId)`, `stopMediaRelay(meetingId)`	The parameter is `destinationRoomId`.
`requestMediaRelay(..., kinds: ['audio', 'video'])`	`kinds: [StreamKind.AUDIO, StreamKind.VIDEO]`. The values are `AUDIO`, `VIDEO` and `SHARE`; `share_audio` has no replacement.
`pauseAllStreams(kind: 'video')`, `resumeAllStreams(kind: ...)`	`kind: StreamKind.VIDEO`. The values are `AUDIO`, `VIDEO` and `SHARE`.

`Participant`

v4.0.1	v5.0.0
`unmuteMic()`, `muteMic()` on a remote participant	`enableMic()`, `disableMic()`
`quality` (`String?`)	`videoQuality` (`VideoQuality?`)
`setQuality("high")`	`setQuality(VideoQuality.HIGH)`. The values are `HIGH`, `MEDIUM` and `LOW`.
`isAgent`	`isAgent()`
`AgentParticipant.agentId`, `agentState`	`getAgentId()`, `getAgentState()`

Types and enums

v4.0.1	v5.0.0
`Stream`	`RTCStream`. `kind` is a `MediaType` (`VIDEO`, `AUDIO`, `SHARE` or `SHARE_AUDIO`), and `participantId` says whose stream it is.
`CustomTrack`, `CustomVideoTrackConfig`, `CustomAudioTrackConfig`	`CustomStreamTrack`, `CustomCameraTrackConfig`, `CustomMicrophoneTrackConfig` (section 3)
`TranscriptionConfig(webhookUrl:, summaryConfig:)`	`TranscriptionConfig(webhookURL:, summary:)`, plus the new `modelConfig`
`PostTranscriptionConfig(summaryConfig:)`	`PostTranscriptionConfig(summary:)`. `SummaryConfig.prompt` is optional.
`E2EEState.EncryptionSucessed`, `DecryptionSucessed`	`E2EEState.EncryptionSuccess`, `DecryptionSuccess`
`TranslationText.type` (`String?`)	`TranslationType?`: `PARTIAL`, `FULL` or `null`
`'recording': {'enabled': true}` in a livestream config map	`LivestreamConfig(recording: LivestreamRecordingConfig(enabled: true))`
`FacingMode.user`, `FacingMode.environment`	`FacingMode.FRONT`, `FacingMode.BACK`
Lower-case values of `RoomState`, `LogLevel`, `LeaveReason`, `AgentState` and `CaptureFlash`, such as `RoomState.connected`	Upper case, such as `RoomState.CONNECTED`, `LogLevel.INFO`, `LeaveReason.MANUAL_LEAVE_CALLED`, `AgentState.SPEAKING` and `CaptureFlash.ON`. `LeaveReason.DUPLICATE_PARTICIPANT` is new.

Events

v4.0.1	v5.0.0
`liveStreamStarted`, `liveStreamStopped`, `liveStreamStateChanged`	`livestreamStarted`, `livestreamStopped`, `onLivestreamStateChanged`
`e2eeStateChanged`	`onE2EEStateChanged`, on the room and on the participant (section 10)
`hlsStarted`, `hlsStopped`	Removed. Use `hlsStateChanged`.
`hlsStateChanged`, `(Map data)`	`(String state, HLSUrl? hlsUrl)`
`participantLeft`, `(String participantId, LeaveReason reason)`	`(Participant participant, LeaveReason reason)`, delivered once per departure
`participantModeChanged` map key `mode` (`String`)	`Mode`
`pinStateChanged` map key `state` (a `{cam, share}` map)	`ParticipantPinState`, with `cam` and `share`
`micRequested`, `cameraRequested` map	Adds `participantId`, who sent the request
`streamStateChanged` on a stream, `(String state, int timestamp)`	`(StreamState state, int timestamp)`: `StreamState.ACTIVE`, `ENDED`, `STUCK`, `FREEZE_DETECTED` or `FREEZE_RESOLVED`
`transcriptionStateChanged`, `(Map data)`	`(TranscriptionState state)`
`translationLanguageChanged`, `(String language)`	`(String? participantId, String language)`
`error`, `(Map error)`	`(VideoSDKError error)` (section 11)
`entryResponded` map key `id`	`participantId`
`roomJoined` after `switchTo()`, map key `switchingRoomId`	`switchRoomId`
`mediaRelayRequestReceived`, `(meetingId, peerId, displayName, accept, reject)`	`(participantId, roomId, displayName, accept, reject)`. The first two arguments swapped places.
`mediaRelayStarted`, `mediaRelayStopped`, `mediaRelayError` with a `meetingId`	The same arguments. The first one is now named `roomId`.
`mediaRelayRequestResponse`, `decision` (`String`)	`RelayDecision?`: `ACCEPTED` or `REJECTED`
`pausedAllStreams`, `resumedAllStreams`, `(String kind)`	`(StreamKind? kind)`, `null` when every kind was paused or resumed
`VideoSDK.on(Events.deviceChanged, ...)`	`room.setAudioDeviceChangeListener((selectedDevice, devices) {...})`

Other signature changes

v4.0.1	v5.0.0
A `switch` over `Events` or `CustomScreenShareTrackConfig` with no `default` case	Add cases for the new events (`dataReceived`, `videoEffectsStats`, `externalCallRinging`, `externalCallAnswered`, `externalCallRejected` and `externalCallHangup`) and for `CustomScreenShareTrackConfig.h480p_15fps`, `h480p_30fps` and `h720p_30fps`, or add a `default`.
A class that extends or implements `Participant`, such as a test mock	Update the overrides to the new names above. `captureImage()` gained `CaptureFlash flash = CaptureFlash.OFF`, and the actions listed in section 8 return `Future<void>`.
A class that implements `Room`, such as a test fake	Update the overrides to the new names above. `end()`, `stopTranslation()` and the undocumented `setViewPort()` return `Future<void>`, and `enableScreenShare()` takes an optional `customTrack`.
`Events.userMessage`, `UserMessage` and `CustomTrack.toMap()` (undocumented)	Removed, with no replacement.

7. Native video processors

Android: a custom processor now implements the SDK's VideoFrameProcessor instead of VideoProcessor from videosdk_webrtc. The method is onFrameCaptured(), and VideoFrame comes from the SDK.

Before

YourOwnBackgroundProcessor.kt
import live.videosdk.webrtc.VideoProcessor
import org.webrtc.VideoFrame

class YourOwnBackgroundProcessor : VideoProcessor {
    override fun onFrameReceived(videoFrame: VideoFrame): VideoFrame {
        return applyEffect(videoFrame)
    }
}

After

YourOwnBackgroundProcessor.kt
import live.videosdk.rtc.android.listeners.VideoFrameProcessor
import videosdk.org.webrtc.VideoFrame

class YourOwnBackgroundProcessor : VideoFrameProcessor {
    override fun onFrameCaptured(videoFrame: VideoFrame): VideoFrame {
        return applyEffect(videoFrame)
    }
}

Key points:

Register the processor in MainActivity with VideoSDK.getInstance().registerVideoProcessor() as before. Apply it from Dart with room.setVideoProcessor(name) instead of VideoSDK.applyVideoProcessor(videoProcessorName:), and remove it with room.removeVideoProcessor(). Both work before join(). A name that was never registered makes the call throw.
VideoSDK.getInstance().setVideoProcessor(processor) in Kotlin applies a processor to the current room directly. It throws IllegalStateException until Dart has called createRoom().
VideoFrame now ships with the SDK as videosdk.org.webrtc.VideoFrame, so the processor no longer needs the io.github.webrtc-sdk:android dependency. Remove it from your app's build.gradle.
The SDK passes I420 frames and releases both the frame it passes in and the frame you return. Do not call release() on either.
iOS: custom processors are not supported in v5.0.0. The VideoSDK.getInstance.registerVideoProcessor() registry is gone, so remove that code from your AppDelegate, together with the processor class built on videosdk_webrtc.
iOS: room.setVideoProcessor(name) reads the name as the file path or URL of a background image, and replaces the camera background with that image.
For background blur on both platforms, use VideoSDK.setVideoEffects().

8. Actions return a `Future` that throws `VideoSDKError`

In v4.0.1, an action that could not run, for example before join or while the room reconnected, reported the failure on Events.error and returned normally. In v5.0.0, Room, Participant and RTCStream actions return a Future that throws VideoSDKError when the SDK refuses them. On Android, most actions called while the room reconnects still report 3065 on Events.error and return without throwing, so keep your Events.error handler.

This compiles without a warning. A call you do not await turns a refusal into an unhandled async error, which crash reporters record as a crash.

Before

void onEndCallPressed(Room room) {
  room.end();
}

void onPinPressed(Participant participant) {
  participant.pin();
}

After

Future<void> onEndCallPressed(Room room) async {
  try {
    await room.end();
  } on VideoSDKError catch (error) {
    log("Could not end the meeting: ${error.code} ${error.name}");
  }
}

Future<void> onPinPressed(Participant participant) async {
  try {
    await participant.pin();
  } on VideoSDKError catch (error) {
    log("Could not pin: ${error.code} ${error.name}");
  }
}

These members returned void or dynamic in v4.0.1 and now return Future<void>:

Class	Members
`Room`	`end()`, `stopTranslation()`
`Participant`	`disableMic()` and `enableMic()` (were `muteMic()` and `unmuteMic()`), `enableCam()`, `disableCam()`, `pin()`, `unpin()`, `remove()`, `setQuality()`, `setViewPort()`, `consumeMicStreams()`, `stopConsumingMicStreams()`, `consumeWebcamStreams()`, `stopConsumingWebcamStreams()`
`RTCStream` (was `Stream`)	`pause()`

Key points:

The other Room actions, such as muteMic(), enableCam() and startRecording(), already returned a Future in v4.0.1. They can now throw too.
Common codes: 3054 when there is nothing to do, for example muteMic() on a muted microphone, 3055 when the same action is already running, 3065 while the room reconnects, and 3035 for an action after the room has left.
Turn on the unawaited_futures and discarded_futures lints in analysis_options.yaml to find calls you do not await. unawaited_futures checks async functions, and discarded_futures checks other functions, such as the Before handlers above. Neither flags an arrow callback such as onPressed: () => room.end(), so check those by hand. discarded_futures also flags other Future calls, so you may want it on only while you migrate.
Events.error also reports some failures that a call of yours caused, such as a failed room.send() (3070), so the same failure can reach both your catch and your Events.error handler. Handle each one in one place.
AgentParticipant changed the same way: pin(), unpin() and remove() return Future<void>.

9. Joining and leaving a room

join() now completes once the room is joined and throws when it is not. A room is single-use: after Events.roomLeft, or after a failed join(), it removes your handlers and releases itself.

Before

Future<void> startMeeting(Room room) async {
  room.on(Events.error, (error) {
    log("Meeting error: ${error['code']} ${error['message']}");
  });
  await room.join();
}

void endCall(Room room) {
  room.leave();
}

After

Future<void> startMeeting(Room room) async {
  room.on(Events.error, (VideoSDKError error) {
    log("Meeting error: ${error.code} ${error.message}");
  });
  // On Android, join() no longer asks for camera or microphone permission.
  await VideoSDK.requestPermissions();
  try {
    await room.join();
  } on VideoSDKError catch (error) {
    // For example 4002 (invalid token) or 3073 (empty room id).
    log("Join failed: ${error.code} ${error.name}");
  }
}

Future<void> endCall(Room room) async {
  try {
    await room.leave();
  } on VideoSDKError catch (error) {
    log("Leave failed: ${error.code} ${error.name}");
  }
}

Key points:

In v4.0.1, join() reported failures on Events.error. In v5.0.0 it throws them.
On Android, join() and changeMode(Mode.SEND_AND_RECV) no longer ask for camera or microphone permission, and without it the microphone and camera stay off. Call VideoSDK.requestPermissions() first (section 3).
To join again, create a new Room with VideoSDK.createRoom(). Calling join() on a room that has left throws a StateError.
Calling join() again on the same room throws VideoSDKError 3055 while the first join is still running, and 3054 once the room is joined.
Only one Room exists at a time. A new createRoom() releases the previous room, and leaves its room if it had joined.
If your screen goes away while join() is still running, wait for it and then leave: room.join().then((_) => room.leave()).ignore();.
There is no dispose() to call. Remove any cleanup that disposed the room.
room.localParticipant exists from createRoom(), as in v4.0.1. Before join it carries the id and name you passed, empty when you passed none; join fills in the ones the room assigned.

10. Stream events also arrive on the `Room`

Handlers added with participant.on() for streamEnabled, streamDisabled, streamPaused and streamResumed still work, and so do e2eeStateChanged handlers once you rename the event to onE2EEStateChanged. The stream type is now RTCStream. The Room now also delivers each of these events, for every participant.

Before

@override
void initState() {
  super.initState();
  widget.participant.on(Events.streamEnabled, (Stream stream) {
    if (stream.kind == 'video') {
      setState(() => videoStream = stream);
    }
  });
}

After

@override
void initState() {
  super.initState();
  widget.participant.on(Events.streamEnabled, (RTCStream stream) {
    if (stream.kind == MediaType.VIDEO) {
      setState(() => videoStream = stream);
    }
  });
}

Event	On the participant	On the room (new)
`streamEnabled`, `streamDisabled`	`(RTCStream stream)`	`(RTCStream stream)`. `stream.participantId` says whose stream it is.
`streamPaused`, `streamResumed`	`(String? kind, String? reason)`	`(String? kind, String? reason)`. It does not say whose stream it is.
`onE2EEStateChanged`	`(E2EEState state, RTCStream? stream)`	`(Map<String, dynamic> data)`, with `participantId`, `streamId` and `state`, an `E2EEState`

Key points:

Take room handlers off with room.off() when the widget that added them goes away. The room outlives a participant tile.
participant.streams is still kept up to date, so a widget can pick up the streams that already exist when it is built.

11. Error codes: `code` is an `int`, and 3xxx codes were renumbered

Events.error now delivers a VideoSDKError instead of a Map. Its code is an int, where v4.0.1 sent a String. Errors thrown by actions are the same VideoSDKError, with code, name and message. A handler declared as (error) with no type still compiles, and error['code'] then fails at runtime, so give the parameter its type. Most 3xxx numbers now mean a different error.

Before

room.on(Events.error, (error) {
  if (error['code'] == "4009") {
    showRoomFullDialog();
  }
});
room.join();

After

room.on(Events.error, (VideoSDKError error) {
  final code = error.code; // an int, for example 4009
  log("Meeting error: $code ${error.name}");
});

try {
  await room.join();
} on VideoSDKError catch (error) {
  // A full room fails join() itself. Show the dialog only here, because the
  // same failure can also reach Events.error.
  if (error.code == 4009) showRoomFullDialog();
}

Same error, new code. Update any check on these numbers:

Error	v4.0.1	v5.0.0
`ERROR_CAMERA_IN_USE`	`3003`	`3023`
`ERROR_MICROPHONE_IN_USE`	`3004`	`3024`
`ERROR_CAMERA_ACCESS_UNAVAILABLE`	`3005`	`3033`
`ERROR_MICROPHONE_ACCESS_UNAVAILABLE`	`3006`	`3034`
`ERROR_CAMERA_ACCESS_DENIED_OR_DISMISSED`	`3007`	`3017`
`ERROR_MICROPHONE_ACCESS_DENIED_OR_DISMISSED`	`3008`	`3018`
`ERROR_CAMERA_PERMISSION_DENIED_BY_OS`	`3009`	`3019`
`ERROR_MICROPHONE_PERMISSION_DENIED_BY_OS`	`3010`	`3020`
`ERROR_ACTION_PERFORMED_BEFORE_MEETING_JOINED`, now `ERROR_ACTION_PERFORMED_BEFORE_ROOM_JOINED`	`3022`	`3035`
`ERROR_CHANGE_WEBCAM`	`3024`	Not reported. A failed `changeCam()` throws another code, for example `3051` for an unknown camera.
`ERROR_VIDEO_PRODUCE_CODEC_NOT_SUPPORTED`	`3025`	`3045`
`ERROR_MULTISTREAM_NOT_SUPPORTED`	`3026`	`3046`
`ERROR_MEETING_RECONNECTING`, now `ERROR_ROOM_RECONNECTING`	`3027`	`3065`

Same code, new meaning. A v4.0.1 check on these numbers now matches a different error:

Code	v4.0.1	v5.0.0
`3010`	`ERROR_MICROPHONE_PERMISSION_DENIED_BY_OS`	`ERROR_BLUETOOTH_AUDIO_ROUTE_FAILED` (Android only)
`3011`	`ERROR_STARTING_VIDEO`	`ERROR_GET_VIDEO_MEDIA`
`3012`	`ERROR_STARTING_AUDIO`	`ERROR_GET_AUDIO_MEDIA`
`3013`	`ERROR_STARTING_SCREENSHARE`	`ERROR_GET_DISPLAY_MEDIA`
`3014`	`ERROR_SCREENSHARE_PERMISSION_DENIED`	`ERROR_GET_VIDEO_MEDIA_PERMISSION_DENIED`
`3015`	`ERROR_SCREENSHARE_NOT_SUPPORTED`	`ERROR_GET_AUDIO_MEDIA_PERMISSION_DENIED`
`3016`	`ERROR_STARTING_FOREGROUND_SERVICE`	`ERROR_GET_DISPLAY_MEDIA_PERMISSION_DENIED`
`3017`	`ERROR_CREATING_VIDEO_TRACK`	`ERROR_CAMERA_ACCESS_DENIED_OR_DISMISSED`
`3018`	`ERROR_CREATING_AUDIO_TRACK`	`ERROR_MICROPHONE_ACCESS_DENIED_OR_DISMISSED`
`3019`	`ERROR_CAMERA_DEVICE_INCOMPATIBLE`	`ERROR_CAMERA_PERMISSION_DENIED_BY_OS`
`3020`	`ERROR_MICROPHONE_DEVICE_INCOMPATIBLE`	`ERROR_MICROPHONE_PERMISSION_DENIED_BY_OS`
`3021`	`ERROR_CHANGING_MODE`	`ERROR_CAMERA_NOT_FOUND`
`3022`	`ERROR_ACTION_PERFORMED_BEFORE_MEETING_JOINED`	`ERROR_MICROPHONE_NOT_FOUND`
`3023`	`ERROR_JOINING_MEETING`	`ERROR_CAMERA_IN_USE`
`3024`	`ERROR_CHANGE_WEBCAM`	`ERROR_MICROPHONE_IN_USE`
`3025`	`ERROR_VIDEO_PRODUCE_CODEC_NOT_SUPPORTED`	`ERROR_CAMERA_PERMISSION_OR_AUTOPLAY_ISSUE`
`3026`	`ERROR_MULTISTREAM_NOT_SUPPORTED`	`ERROR_VIDEO_SOURCE_INITIATION_FAILED`
`3027`	`ERROR_MEETING_RECONNECTING`	`ERROR_WEBCAM_TRACK_ENDED`

3001 to 3009 are not reported in v5.0.0.

No longer reported. These v4.0.1 errors have no v5.0.0 code:

v4.0.1 code	Error
`3001`	`ERROR_CUSTOM_VIDEO_TRACK_DISPOSED`
`3002`	`ERROR_CUSTOM_AUDIO_TRACK_DISPOSED`
`3011`	`ERROR_STARTING_VIDEO`
`3012`	`ERROR_STARTING_AUDIO`
`3013`	`ERROR_STARTING_SCREENSHARE`
`3014`	`ERROR_SCREENSHARE_PERMISSION_DENIED`
`3015`	`ERROR_SCREENSHARE_NOT_SUPPORTED`
`3016`	`ERROR_STARTING_FOREGROUND_SERVICE`
`3017`	`ERROR_CREATING_VIDEO_TRACK`
`3018`	`ERROR_CREATING_AUDIO_TRACK`
`3019`	`ERROR_CAMERA_DEVICE_INCOMPATIBLE`
`3020`	`ERROR_MICROPHONE_DEVICE_INCOMPATIBLE`
`3021`	`ERROR_CHANGING_MODE`
`3023`	`ERROR_JOINING_MEETING`

Codes you will see most often in v5.0.0:

Code	Name	Meaning
`3035`	`ERROR_ACTION_PERFORMED_BEFORE_ROOM_JOINED`	The call needs a joined room.
`3051`	`ERROR_NO_WEBCAM_DEVICE_FOUND`	No camera was found.
`3054`	`ERROR_ALREADY_IN_REQUESTED_STATE`	Nothing to do: it is already in that state.
`3055`	`ERROR_OPERATION_IN_PROGRESS`	The same operation is already running.
`3056`	`ERROR_OPERATION_ABORTED`	The operation was aborted.
`3059`	`ERROR_NO_MICROPHONE_DEVICE_FOUND`	No microphone was found.
`3065`	`ERROR_ROOM_RECONNECTING`	The room is reconnecting. Try again once it is back.
`3071`	`ERROR_INVALID_CUSTOM_TRACK_CONFIG`	The track configuration is not one the SDK knows.
`3073`	`ERROR_INVALID_PARAMETER`	An argument is invalid, for example an empty room id.
`4000`	`UNKNOWN_ERROR`	The error could not be classified. The original message is kept.
`4002`	`INVALID_TOKEN`	The token was rejected.

Key points:

Other 4xxx and 5xxx numbers are unchanged. Two names now use the catalogue's spelling: 4009 is MAX_PARTCIPANT_REACHED, where v4.0.1 sent MAX_PARTICIPANT_REACHED, and 4023 is UNAUTHORIZED_PARTCIPANT_ID, where v4.0.1 sent UNAUTHORIZED_PARTICIPANT_ID. Compare these two by code, not by name.
The VideoSDKErrors map is removed. Compare with the VideoSDKError.* constants instead, such as VideoSDKError.INVALID_TOKEN: errors are equal when their code is equal.
Error names say ROOM where v4.0.1 said MEETING: ERROR_ACTION_PERFORMED_BEFORE_ROOM_JOINED, ERROR_ROOM_RECONNECTING, ERROR_ROOM_MEDIA_CONNECTION_FAILED, ERROR_UNABLE_TO_JOIN_ROOM, ERROR_END_ROOM_FAILED, INVALID_ROOM_ID, UNAUTHORIZED_ROOM_ID, ROOM_ID_DISABLED and SWITCH_ROOM_FAILED. The codes are unchanged, so a check on error.name must change and a check on code does not.

12. Other behaviour changes

What	v4.0.1	v5.0.0
`participant.mode`	Updated when the participant changed mode	Updated just before `Events.participantModeChanged` fires, for the local participant too. `changeMode()` does not change it when it returns: wait for the event.
`Events.qualityLimitation` values	Lower case, such as `cpu` and `detected`	Upper case: the type is `CONGESTION`, `BANDWIDTH` or `CPU`, and the state is `DETECTED` or `RESOLVED`. The handler is `(String type, String state, int? timestamp)`. Update any string comparison.
An error thrown in a `room.on()` handler	For some events, such as `Events.participantLeft`, the SDK caught it and only logged it	Reaches your app as an uncaught error, which crash reporters record. Check that each handler's parameter types match the event: `Events.participantLeft` passes `(Participant participant, LeaveReason reason)`, so a handler that declares a `String` first parameter throws. The other handlers for the event still run.
`getVideoStats()`, `getAudioStats()`, `getShareStats()`	Returned the latest sample from stats the SDK collected in the background, or an empty list before the first one	Return the sample fetched by your previous call and start fetching the next one. The first call after a participant joins returns `null`, so call them from a timer.
`room.changeMic()` on Android and iOS	Threw `UnsupportedError` once joined	Works. A device picked before join is used when you join. It also moves the speaker route, because the phone has a single audio route, and `VideoSDK.getSelectedAudioDevice()` returns it. `setAudioDevice()` changes the route without touching the microphone.
Ids and labels from `VideoSDK.getAudioDevices()`, now `getMics()`	Android ids `speaker`, `earpiece`, `headset` and `bluetooth`, with labels such as `Speakerphone`. iOS ids were the audio port's UID	Android ids `SPEAKER_PHONE`, `EARPIECE`, `WIRED_HEADSET` and `BLUETOOTH`, and the label is the same string. On iOS the id is the same string as the label. Update code that picks a device by id or label, and show your own names in device pickers.
Ids and labels from `VideoSDK.getVideoDevices()`, now `getCameras()`	Android labels such as `Camera 1, Facing front, Orientation 270`. iOS ids were the camera's unique id	Android labels such as `Camera 1 (Front Camera)`. On iOS the id is the camera's name.
An unknown `cameraId` or `microphoneId` in `createCameraVideoTrack()` or `createMicrophoneAudioTrack()`	Used the default device	Throws `3071`. A device id saved by a v4.0.1 build may not match, so pick from a fresh `getCameras()` or `getMics()` list.

13. Removed or no longer delivered

v4.0.1	v5.0.0
Web, macOS and Windows support	Not supported. Stay on v4.0.1 for those platforms.
`Events.videoQualityChanged`	Not delivered. No replacement.
`VideoSDK.on(Events.deviceChanged)`	Removed. Use `room.setAudioDeviceChangeListener((selectedDevice, devices) {...})`, or call `VideoSDK.getDevices()` when you need the current list.
`VideoSDK.on(Events.error)` for track errors	Not delivered. Catch `VideoSDKError` from the track factories (section 3), and listen with `room.on(Events.error)` for room errors.
`createRoom(debugMode:)`, `createRoom(defaultCameraIndex:)`	Removed: `debugMode` has no replacement (`VideoSDK.setLogLevel()` sets how much the SDK logs); use `createRoom(defaultCamera:)` for `defaultCameraIndex` (section 2)
`enableScreenShare(enableAudio:, source:)`, `Room.getScreenShareSources()`	`createScreenShareVideoTrack(withAudio:)`. Desktop source picking has no replacement (section 4).
`PubSubException`, `PubSubMeetingNotJoined`, `PubSubMeetingReconnecting`, `PubSubPublishFailed`, `PubSubSubscribeFailed`, `PubSubUnsubscribeFailed`	`VideoSDKError`
`PubSubMessage.sendOnly` on received messages	No replacement
`Mode.CONFERENCE`, `Mode.VIEWER`	Removed. Use `Mode.SEND_AND_RECV` or `Mode.RECV_ONLY`.
`createRoom(notification:)`, `NotificationInfo`	Removed. On Android, the SDK shows its own screen-share notification.
`startLivestream(postTranscriptionConfig:)`	Removed. Transcribe a recording or an HLS stream with `transcription:` instead.
`Events.hlsStarted`, `Events.hlsStopped`	Removed. Use `Events.hlsStateChanged`.
`room.enableRelayMedia()`, `room.disableRelayMedia()`	Removed, with no replacement.
`Character`, `room.createCharacter()` and the `character*` events (undocumented)	Removed, with no replacement.
Custom video processors on iOS	No replacement (section 7)
`videosdk_media_effects` plugin	Noise suppression: `createMicrophoneAudioTrack(noiseConfig:)`. Background image: no replacement on Android; on iOS, `room.setVideoProcessor()` with the image path or URL.
`MediaDeviceType`, `VideoSDK.mediaDevices`, `loadMediaDevices()`, `isFirefox()`, `setAppleAudioConfiguration()`, `getDeviceInfo()`, `requestIOSScreenSharePermission()`, `e2eeType` (undocumented)	Removed. Use `VideoSDK.getCameras()`, `getMics()` and `getDevices()` for devices.
`Room.localParticipantId`, `getParticipentId()`, `switchingRoom()`, `isSendAndReceiveMode()`, `isViewerMode()`, `deviceInfo`, `baseUrl`, `videoSDKTelemetery` (undocumented)	Removed. Use `room.localParticipant.id` and `participant.mode`.
`Room.enableShare()`, `disableShare()`, `pause(consumer)`, `resume(consumer)`, `setConsumerQuality()`, `setScreenShareQuality()` and `setViewPort()` on the `Room` (undocumented)	Removed. Use `enableScreenShare()` and `disableScreenShare()`, `stream.pause()` and `stream.resume()`, and the `Participant` methods `setQuality()`, `setScreenShareQuality()` and `setViewPort()`.
`RTCStream.track`, `copyWith()`, `RealtimeStore.dispose()`, `RTCVideoRenderer.audioOutput()` (undocumented)	Removed, with no replacement.
`BaseKeyProvider.key`, `sharedKey`, `keyMap`, `keyProvider`	Hidden from the API reference. Use `setSharedKey()` on `BaseKeyProvider` and `VideoSDK.setParticipantKey()` for keys.

14. New in v5.0.0

Feature	API
Pre-call test	`VideoSDK.runPreCallTest(PreCallTestOptions(token: token))` returns a `Future<PreCallTestResult>`. Stop it early with `VideoSDK.stopPreCallTest()`. Run it, and stop it, before `createRoom()`: creating a room does not stop a running test. Failures throw `PreCallTestException`. See Precall.
Video effects	`VideoSDK.setVideoEffects(VideoEffectsConfig(...))` or `room.setVideoEffects()` for framing, low light, style, color filter and background blur, plus `getAppliedVideoEffectsConfig()`, `publishFramingSubject()` and `Events.videoEffectsStats`. On Android, framing on the default `VideoEffectsPipeline.LEGACY` needs `com.google.mlkit:face-detection:16.1.7` and blur needs `com.google.mlkit:segmentation-selfie:16.0.0-beta6` in your app's `build.gradle`. On iOS, background blur needs iOS 15 or later.
Data channel	`room.send()` sends a `String` or a `List<int>` of up to 15 KiB. Receive it with `Events.dataReceived`, which delivers a `DataMessage` with `senderId`, `text`, `data`, `isBinary`, `reliable` and `timestamp`.
Typed errors	`VideoSDKError` with `code`, `name` and `message`
Log forwarding	`VideoSDK.enableLogForwarding((level, message) {...})` and `disableLogForwarding()`
Version info	`VideoSDK.sdkVersion` and `VideoSDK.getBuildId()` for bug reports
Capture flash	`captureImage(flash: CaptureFlash.ON)` and `VideoSDK.setScreenFlash(true)`. Local participant only.
Outgoing video quality	`createRoom(dynaCast: true)` lowers your webcam quality while nobody watches its larger layers, and has no effect with `multiStream: false`. `room.setWebcamQuality(VideoQuality.MEDIUM)` caps it.
Received screen-share quality	`participant.setScreenShareQuality(VideoQuality)`
More statistics	`participant.getTransportStats()` and `participant.getShareAudioStats()`
Applied settings	`room.getAppliedVideoConfig()`
Room attributes	`room.getAttributes()` returns the room's server-side attributes.
Room state	`room.roomState`, `room.isScreenSharing`, `room.transcriptionState`, `room.translationState`
Stream details	`StreamState` for `streamStateChanged`, `MediaType.SHARE_AUDIO` for screen-share audio, `CameraDeviceInfo.facingMode`
Error constants	`VideoSDKError.*`, such as `VideoSDKError.INVALID_TOKEN`; errors are equal by `code`
Device switching	`room.changeCam()` with no arguments moves to the next camera, and `room.changeMic()` now works on Android and iOS. `room.setAudioDeviceChangeListener()` reports route changes.
Entry requests	`room.respondEntry(participantId, 'allowed')` answers a join request from anywhere, not only from the `Events.entryRequested` handler.
Listener cleanup	`room.removeAllEventListeners()` and `participant.removeAllEventListeners()`
Phone calls	`Events.externalCallRinging`, `externalCallAnswered`, `externalCallRejected` and `externalCallHangup`, each with a `CallType`: `INCOMING` or `OUTGOING`
Track end	`CustomStreamTrack.onEnded` is called once when a track ends, with a `VideoSDKError` when the device was lost or access was revoked.
E2EE off	`VideoSDK.removeKeyProvider()` turns end-to-end encryption off. `BaseKeyProvider.removeKey(participantId)` removes one participant's key.
Mirroring	`RTCVideoRenderer.setMirror(bool)` flips every view of a renderer, overriding `RTCVideoView(mirror:)`.
Screen-share profiles	`CustomScreenShareTrackConfig.h480p_15fps`, `h480p_30fps` and `h720p_30fps`
High-pass filter	`highPassFilter` in `NoiseConfig`
Realtime translation	`startTranslation()`, `stopTranslation()` and `changeTranslationLanguage()` are now documented. See Realtime Translation.

API Reference

Got a Question? Ask us on discord

1. Remove videosdk_webrtc​

2. VideoSDK.createRoom() parameters changed​

3. Track factories are typed and throw​

4. Screen share takes a track​

5. PubSub subscribe() takes a listener object​

6. Renamed APIs​

VideoSDK​

Room​

Participant​

Types and enums​

Events​

Other signature changes​

7. Native video processors​

8. Actions return a Future that throws VideoSDKError​

9. Joining and leaving a room​

10. Stream events also arrive on the Room​

11. Error codes: code is an int, and 3xxx codes were renumbered​

12. Other behaviour changes​

13. Removed or no longer delivered​

14. New in v5.0.0​

API Reference​

1. Remove `videosdk_webrtc`

2. `VideoSDK.createRoom()` parameters changed

3. Track factories are typed and throw

4. Screen share takes a track

5. PubSub `subscribe()` takes a listener object

6. Renamed APIs

`VideoSDK`

`Room`

`Participant`

Types and enums

Events

Other signature changes

7. Native video processors

8. Actions return a `Future` that throws `VideoSDKError`

9. Joining and leaving a room

10. Stream events also arrive on the `Room`

11. Error codes: `code` is an `int`, and 3xxx codes were renumbered

12. Other behaviour changes

13. Removed or no longer delivered

14. New in v5.0.0

API Reference