Can't reuse PeerConnectionClient after activity recreated - android

I'm trying to correctly handle screen rotation during WebRTC call on Android.
But after first rotation local video translation stopping.
After creating (or recreating) activty I am creating SurfaceViewRenderers for local & remote views:
ourView.init(eglBase.eglBaseContext, null)
ourView.setScalingType(RendererCommon.ScalingType.SCALE_ASPECT_FIT)
ourView.setZOrderMediaOverlay(true)
ourView.setEnableHardwareScaler(true)
ourView.setMirror(true)
theirView.init(eglBase.eglBaseContext, null)
theirView.setScalingType(RendererCommon.ScalingType.SCALE_ASPECT_FIT)
theirView.setEnableHardwareScaler(false)
theirView.setMirror(true)
localVideoSyncer.setTarget(ourView)
remoteVideoSyncer.setTarget(theirView)
After that, if this is a first time when connection must be created, I am initializing peer connection:
iceConnected = false
val dataChannelParameters = CallUtils.createChannelParameters()
peerConnectionParameters = CallUtils.createConnectionParameters(videoWidth, videoHeight, dataChannelParameters)
if(appRTCClient == null) {
appRTCClient = CallUtils.createAppRTCClient(roomId, this, info?.ice ?: emptyList())
}
roomConnectionParameters = CallUtils.createRoomConnectionParameters(info?.address, roomId)
if(peerConnectionClient == null) {
peerConnectionClient = PeerConnectionClient(
applicationContext, eglBase, peerConnectionParameters, this)
}
val options = PeerConnectionFactory.Options()
peerConnectionClient!!.createPeerConnectionFactory(options)
if (appRTCClient == null) {
NetworkLogger.log(TAG,"AppRTC client is not allocated for a call.")
return
}
if(callStartedTimeMs == 0L)
callStartedTimeMs = System.currentTimeMillis()
appRTCClient!!.connectToRoom(roomConnectionParameters!!)
if(audioManager == null)
audioManager = AppRTCAudioManager(applicationContext)
If activity is recreated, PeerConnectionClient is no longer can send video to remote target. I tried to reassign localRender & videoCapturer, but that has no effect. Is it possible to reuse existing connection after activity recreated?

I think the best solution here is to place appRTCClient and related objects outside of activity and make them independent from activity's lifecycle. You use applicationContext thats why you can store it in some singletone which relies on application's lifecycle. Another way is to use fragment inside your activity for displaying chat windows and manually handle rotatation inside onConfigurationChanged method, so you will have to replace portrait fragment with landscape one.

Related

On Android, how to check the phone screen is casting?

I need to check if the Android phone my app runs on is using casting which is enabled outside of my app.
It seems CastSession or SessionManager can provide the session related to my app which is not helpful for me.
For example, I can start casting with an app called xx which will cast or mirror the entire screen of my phone. Now, I need to notify when I open my app that the phone's screen is casting/mirroring so I can prevent showing specific content on my app.
I checked it with the code below:
val isCastingEnabledLiveData = MutableLiveData<Boolean>()
fun isCastingEnabled(context: Context): Boolean {
val mediaRouter = MediaRouter.getInstance(context)
if (mediaRouter.routes.size <= 1) {
isCastingEnabledLiveData.value = false
return
}
val selector = MediaRouteSelector.Builder()
.addControlCategory(MediaControlIntent.CATEGORY_LIVE_VIDEO)
.addControlCategory(MediaControlIntent.CATEGORY_REMOTE_PLAYBACK)
.build()
mediaRouter.addCallback(selector, object : MediaRouter.Callback() {
override fun onRouteChanged(router: MediaRouter?, route: MediaRouter.RouteInfo?) {
super.onRouteChanged(router, route)
isCastingEnabledLiveData.value = if (route != mediaRouter.defaultRoute) {
route?.connectionState != MediaRouter.RouteInfo.CONNECTION_STATE_DISCONNECTED
} else false
}
})
}
you can check whether the phone screen is casting or not by using the MediaRouter class.
Here is an example of how you could check if the phone screen is casting:
MediaRouter mediaRouter = (MediaRouter)
getSystemService(Context.MEDIA_ROUTER_SERVICE);
MediaRouter.RouteInfo route = mediaRouter.getSelectedRoute();
if(route.isDefault()){
// Screen is not casting
} else {
// Screen is casting
}
This code uses the getSelectedRoute() method of the MediaRouter class to get the currently selected route. If the returned RouteInfo object is the default route, then the screen is not casting, otherwise it is.
Please note that this code uses the getSystemService(Context.MEDIA_ROUTER_SERVICE) method to get an instance of the MediaRouter class, so it should be called from an Activity or Service context.
Additionally, you could also use MediaRouter.Callback and MediaRouter.addCallback to set a callback to monitor the state of the casting, so you could get the updates on the casting state change as well.

How to implement back button handling on Android with Redux

I'm trying to implement back button handling on Android using CoRedux for my Redux store. I did find one way to do it, but I am hoping there is a more elegant solution because mine seems like a hack.
Problem
At the heart of the problem is the fact returning to an Android Fragment is not the same as rendering that Fragment for the first time.
The first time a user visits the Fragment, I render it with the FragmentManager as a transaction, adding a back stack entry for the "main" screen
fragmentManager?.beginTransaction()
?.add(R.id.myFragmentContainer, MyFragment1())
?.addToBackStack("main")?.commit()
When the user returns to that fragment from another fragment, the way to get back to it is to pop the back stack:
fragmentManager?.popBackStack()
This seems to conflict with Redux principles wherein the state should be enough to render the UI but in this case the path TO the state also matters.
Hack Solution
I'm hoping someone can improve on this solution, but I managed to solve this problem by introducing some state that resides outside of Redux, a boolean called skipRendering. You could call this "ephemeral" state perhaps. Initialized to false, skipRendering gets set to true when the user taps the back button:
fun popBackStack() {
fragmentManager?.popBackStack()
mapViewModel.dispatchAction(MapViewModel.ReduxAction.BackButton)
skipRendering = true
}
Dispatching the back button to the redux store rewinds the redux state to the prior state as follows:
return when (action) {
// ...
ReduxAction.BackButton -> {
state.pastState
?: throw IllegalStateException("More back taps processed than past state frames")
}
}
For what it's worth, pastState gets populated by the reducer whenever the user requests to visit a fragment from which the user can subsequently tap back.
return when (action) {
// ...
ReduxAction.ShowMyFragment1 -> {
state.copy(pastState = state, screenDisplayed = C)
}
}
Finally, the render skips processing if skipRendering since the necessary work of calling fragmentManager?.popBackStack() was handled before dispatching the BackButton action.
I suspect there is a better solution which uses Redux constructs for example a side effect. But I'm stuck figuring out a way to solve this more elegantly.
To solve this problem, I decided to accept that the conflict cannot be resolved directly. The conflict is between Redux and Android's native back button handling because Redux needs to be master of the state but Android holds the back stack information. Recognizing that these two don't mix well, I decided to ditch Android's back stack handling and implement it entirely on my Redux store.
data class LLReduxState(
// ...
val screenBackStack: List<ScreenDisplayed> = listOf(ScreenDisplayed.MainScreen)
)
sealed class ScreenDisplayed {
object MainScreen : ScreenDisplayed()
object AScreen : ScreenDisplayed()
object BScreen : ScreenDisplayed()
object CScreen : ScreenDisplayed()
}
Here's what the reducer looks like:
private fun reducer(state: LLReduxState, action: ReduxAction): LLReduxState {
return when (action) {
// ...
ReduxAction.BackButton -> {
state.copy(screenBackStack = mutableListOf<ScreenDisplayed>().also {
it.addAll(state.screenBackStack)
it.removeAt(0)
})
}
ReduxAction.AButton -> {
state.copy(screenBackStack = mutableListOf<ScreenDisplayed>().also {
it.add(ScreenDisplayed.AScreen)
it.addAll(state.screenBackStack)
})
}
ReduxAction.BButton -> {
state.copy(screenBackStack = mutableListOf<ScreenDisplayed>().also {
it.add(ScreenDisplayed.BScreen)
it.addAll(state.screenBackStack)
})
}
ReduxAction.CButton -> {
state.copy(screenBackStack = mutableListOf<ScreenDisplayed>().also {
it.add(ScreenDisplayed.CScreen)
it.addAll(state.screenBackStack)
})
}
}
}
In my fragment, the Activity can call this API I exposed when the Activity's onBackPressed() gets called by the operating system:
fun popBackStack() {
mapViewModel.dispatchAction(MapViewModel.ReduxAction.BackButton)
}
Lastly, the Fragment renders as follows:
private fun render(state: LLReduxState) {
// ...
if (ScreenDisplayed.AScreen == state.screenBackStack[0]) {
fragmentManager?.beginTransaction()
?.replace(R.id.llNavigationFragmentContainer, AFragment())
?.commit()
}
if (ScreenDisplayed.BScreen == state.screenBackStack[0]) {
fragmentManager?.beginTransaction()
?.replace(R.id.llNavigationFragmentContainer, BFragment())
?.commit()
}
if (ScreenDisplayed.CScreen == state.screenBackStack[0]) {
fragmentManager?.beginTransaction()
?.replace(R.id.llNavigationFragmentContainer, CFragment())
?.commit()
}
}
This solution works perfectly for back button handling because it applies Redux in the way it was meant to be applied. As evidence, I was able to write automation tests which mock the back stack as follows by setting the initial state to one with the deepest back stack:
LLReduxState(
screenBackStack = listOf(
ScreenDisplayed.CScreen,
ScreenDisplayed.BScreen,
ScreenDisplayed.AScreen,
ScreenDisplayed.MainScreen
)
)
I've left some details out which are specific to CoRedux.

What does Android use to disallow starting service in background?

I am trying to startService only while in foreground, but to determine foreground I use Activity.onStart, this however is called sometimes when app is in background as per that piece of code that throws java.lang.IllegalStateException: Not allowed to start service Intent when starting the service.
-- concretly, when starting activity with screen off.
Does anyone know what the actual check is? How can I "if" this, i.e. I need to && onStart with some processIsInForeground() to get true onStart
So far I have
val appProcesses = activityManager.runningAppProcesses
if (appProcesses == null) return false
val thisProcess = appProcesses.firstOrNull { it.processName == packageName }
if (thisProcess == null) return false
val isProcessForeground == ActivityManager.RunningAppProcessInfo.IMPORTANCE_FOREGROUND
But this only works like 90% of the time, some times importance is IMPORTANCE_SERVICE, sometimes IMPORTANCE_TOP_SLEEPING
Also, docs say that api is mostly just for debug, etc.

Android LiveData observes stale data after navigating back

Question:
How can I prevent my livedata immediately receiving stale data when navigating backwards? I am using the Event class outlined here which I thought would prevent this.
Problem:
I open the app with a login fragment, and navigate to a registration fragment when a live data email/password is set (and backend call says "this is a new account go register"). If the user hits the back button during the registration Android is popping back to login.
When the login fragment is recreated after a back press, it immediately fires the live data again with the stale backend response and I would like to prevent that.
LoginFragment.kt
override fun onViewCreated(view: View, savedInstanceState: Bundle?) {
super.onViewCreated(view, savedInstanceState)
subscribeToLoginEvent()
}
private fun subscribeToLoginEvent() {
//When a back press occurs, we subscribe again and this instantly
//fires with the same data it used to leave the screen
//(a Resource<User> with status=SUCCESS, data = null)
viewModel.user.observe(viewLifecycleOwner, Observer { response ->
Timber.i("login event observed....status:" + response?.status + ", data: " + response?.data)
binding.userResource = response
response?.let {
val status = it.status
val message = it.message
if (status == Status.SUCCESS && it.data == null) {
//This is a brand new user so we need to register now
navController()
.navigate(LoginFragmentDirections.showUserRegistration()))
}
else if(status == Status.SUCCESS && it.data != null){
goHome()
}
}
})
}
LoginViewModel.kt
private val _loginCredentials: MutableLiveData<Event<Pair<String, String>>> = MutableLiveData()
val user: LiveData<Resource<User>> = Transformations.switchMap(_loginCredentials) {
val data = it.getContentIfNotHandled()
if(data != null && data.first.isNotBlank() && data.second.isNotBlank())
interactor.callUserLoginRepo(data.first, data.second)
else
AbsentLiveData.create()
}
Okay there's two issues here which I hope helps somebody else.
The first is that the Event class does not appear to work with transformations. I think it is because the Event is literally pointing to the wrong live data (_login_credentials vs user)
The second problem is a little bit more fundamental but also blindingly obvious now. We are told all over the place that live data observations emit the latest data when a subscription is made to ensure you get the most up to date data. This means the way I am using live data here is simply incorrect, I can't subscribe to a login event, navigate somewhere, navigate back and re-subscribe because the ViewModel is rightfully giving me the latest data it has (because the login fragment was only detached, never destroyed).
Solution
The solution is to simply move the logic which performs the fetch one fragment deeper. So instead of listening for user credentials + login click -> fetching a user -> and then navigating somewhere, I need to listen for user credentials + login click -> navigate somewhere -> and then start subscribing for my user live data. That way I can head back to the login screen as much as I want and not subscribe to some stale live data that was never destroyed. And if I go back to login and then forwards the subscription and fragment were destroyed so I will appropriately be getting new data in that case.

How to remember state with retry operators in RxJava2

I have a network client that is able to resume from interruptions, but needs the last message for doing so when there is a retry.
Example in Kotlin:
fun requestOrResume(last: Message? = null): Flowable<Message> =
Flowable.create({ emitter ->
val connection = if (last != null)
client.start()
else
client.resumeFrom(last.id)
while (!emitter.isDisposed) {
val msg = connection.nextMessage()
emitter.onNext(msg)
}
}, BackpressureStrategy.MISSING)
requestOrResume()
.retryWhen { it.flatMap { Flowable.timer(5, SECONDS) } }
// how to pass the resume data when there is a retry?
Question: as you can see, I need the last received message in order to prepare the resume call. How can I keep track of it so that when there is a retry it is available to make the resume request?
One possible solution may be to create a holder class that just holds a reference to the last message and is updated when a new message is received. This way when there is a retry the last message can be obtained from the holder. Example:
class MsgHolder(var last: Message? = null)
fun request(): Flowable<Message> {
val holder = MsgHolder()
return Flowable.create({ emitter ->
val connection = if (holder.last != null)
client.start()
else
client.resumeFrom(holder.last.id)
while (!emitter.isDisposed) {
val msg = connection.nextMessage()
holder.last = msg // <-- update holder reference
emitter.onNext(msg)
}
}, BackpressureStrategy.MISSING)
}
I think this might work, but it feels like a hack (thread synchronization issues?).
Is there a better way to keep track of the state so it is available for retries?
Note that, unless you rethrow a wrapper around your last element (not too functionally different from your existing "hack"-ish solution but way uglier imo), no error handling operators can recover the last element without some outside help because they only get access to streams of Throwable. Instead, see if the following recursive approach suits your needs:
fun retryWithLast(seed: Flowable<Message>): Flowable<Message> {
val last$ = seed.last().cache();
return seed.onErrorResumeNext {
it.flatMap {
retryWithLast(last$.flatMap {
requestOrResume(it)
})
}
};
}
retryWithLast(requestOrResume());
The biggest distinction is caching the trailing value from the last attempt in an Observable with cache rather than doing so manually in a value. Note also that the recursion in the error handler means retryWithLast will continue to extend the stream if subsequent attempts continue failing.
Take a close look to buffer() operator: link
You could use it like this:
requestOrResume()
.buffer(2)
From now, your Flowable will return List<Message> with two latests objects

Categories

Resources