I used the replaying behavior in rx to cache the latest emit of database, and scope in rx is .. I guess the duration of FooRepository instance lifetime (foo flow instance lifetime)? Well..semantically its not GlobalScope since FooRepository say only lives during user being logged in -- dagger specifies this If I were to have scope as param that would not be nice api. Or maybe use GlobalScope and just leave it up to dagger to scope the FooRepository object, and therefore GC eats this when it goes out of "scope"? Would that be a leak?

-- also, why is a scope needed if say one wants a

refCount

behavior? I dont see how is it relevant

my flow is

Flow<List<T>>

, can i convert it to

Flow<T>

with something like

flatten

on arrays on std lib?

Is it possible to ensure that a channel only ever receives a single value? For example, setting the capacity to 1, and immediately closing it such that any call to

offer

will throw an exception. I'm worried about the synchronization between the close call and the offer call. ie I need an atomic receiveAndClose (though I don't need it to be a channel)

When you show the exception from a

catch

block in the UI, how do you avoid showing CancellationExceptions? Do you have to add an if check?

Hello, I’m currently trying to determine if my async calls are running concurrently or sequentially. I currently have this code:

val (firstImageBlobName, firstImageThumbnailBlobName, secondImageBlobName, secondImageThumbnailBlobName) = try {
    coroutineScope {
        AsyncImageUploads(
            firstImageBlobName = async {
                fileServiceInterface.uploadImage(
                    ...
                )
            },
            firstImageThumbnailBlobName = async {
                fileServiceInterface.uploadImage(
                    ...
                )
            },
            secondImageBlobName = async {
                fileServiceInterface.uploadImage(
                    ...
                )
            },
            secondImageThumbnailBlobName = async {
                fileServiceInterface.uploadImage(
                    ...
                )
            }
        )
    }
} catch (e: Exception) {
    ....
}

repository.save(
    model.copy(
        firstImage = firstImageRepository.save(
            FirstImage(
                imageURL = firstImageBlobName.await(),
                thumbnailURL = firstImageThumbnailBlobName.await(),
                model = model
            )
        ),
        secondImage = secondImageRepository.save(
            SecondImage(
                imageUrl = secondImageBlobName.await(),
                thumbnailUrl = secondImageThumbnailBlobName.await(),
                model = model
            )
        )
    )
)

The issue is the upload takes around 40 seconds to finish 🤔 so I’m wondering if the async calls are happening concurrently

Hi everybody, I’m having an issue with

BroadcastChannel

and

Flow

. Essentially I’m looking for the

navigate

to behave like rx

PublishSubject

. Issue: since

navigate

flow is created from the channel,

firstOrNull()

on it never finishes. I’m pretty new to coroutines and I will appreciate if somebody can have a look. Thanks!

import io.kotest.matchers.shouldBe
import kotlinx.coroutines.channels.BroadcastChannel
import kotlinx.coroutines.channels.Channel
import kotlinx.coroutines.flow.asFlow
import kotlinx.coroutines.flow.firstOrNull
import kotlinx.coroutines.launch
import kotlinx.coroutines.runBlocking
import org.junit.jupiter.api.Test

class RetainStateTest {

    val navigateChannel = BroadcastChannel<List<String>>(Channel.BUFFERED)
    val navigate = navigateChannel.asFlow()

    val data = listOf("blah", "foo")

    @Test
    fun `should not retain state`() {
        runBlocking {
            val jobPrecondition = launch {
                navigate.firstOrNull() shouldBe data
            }
            navigateChannel.send(data)
            jobPrecondition.join()

            val job = launch {
                navigate.firstOrNull() shouldBe null
            }
            job.join()
        }
    }
}

📣 📣 📣 kotlinx.coroutines 1.4.0 is here! •

StateFlow

and

SharedFlow

are promoted to stable API • Rewritten and fully optimized `Flow.combine`: better performance and no more SIGSEGV on Mediatek devices • Various improvements and bug fixes Full changelog: https://github.com/Kotlin/kotlinx.coroutines/releases/tag/1.4.0

@Vsevolod Tolstopyatov [JB] Could you please clarify if

SendChannel.offer should never throw

was fixed in 1.4.0 release? https://github.com/Kotlin/kotlinx.coroutines/issues/974

Also, do we need separate

-native-mt

library for 1.4.0 release for multi-threaded coroutines on KMM?

What does

mutableSharedFlow.asSharedFlow()

bring over

mutableSharedFlow as SharedFlow

?

I'm using Dispatchers.IO while reading data from files on disk. It can take anywhere from 1-5 mins to complete. But, it's using up basically all of the available resources. Is there a way to kind of cap the amount of work being done on a particular dispatcher?

Is there a coroutine/suspending network library out there yet?

Is there a way to chain

onEach

with

MutableStateFlow

and return a

MutableStateFlow

? My use case -- calling

SavedStateHandle.getLiveData(key: String)

will automatically save new emissions to that live data into the saved state. I would like something like this to be possible with StateFlow

fun <T> SavedStateHandle.getStateFlow(key: String, defaultValue: T): MutableStateFlow<T> {
    return MutableStateFlow(this[key] ?: defaultValue).onEach {
        this[key] = it
    }
}

However,

onEach

returns a regular instance of

Flow

. Is this something that's been discussed? If not, I'm more than willing to open an issue.

When using

BroadcastChannel

in ViewModel in order to send events, I was using the non suspend channel.offer(..) to send them, but now with SharedFlow, there seems only be the suspending

emit(...)

method. Does that mean all BroadcastChannel usage's can be migrated from

events.offer(...)

to Sharedflow?

viewModelScope.launch {
    events.emit(...)
}

If so, that looks like more boilerplate for simple events sending 😕

Coroutines Riddle: What happens when you execute this with coroutines 1.4.0?

inline class A(val value: String)

fun main() {
    runBlocking {
        val aFlow = flowOf(A("Hello"))
        combine(aFlow, aFlow) { a, b ->
            a to b
        }.toList()
    }
}

I have a question about retriggering a flow (Android app with Paging 3)

Hi! I created a “Kotlin Coroutines Exception Handling Cheat Sheet”. It might be helpful 👉 https://www.lukaslechner.com/coroutines-exception-handling-cheat-sheet/

Hey all, so I'm trying to work around an inconsistently slow API. What's the best way to make two async requests but only keep the fastest one to respond? (That infinite

while

seems like an antipattern to me.) Any ideas?

Has anyone here used the new Jetpack DataStore yet? I want to restore the checked state of a checkbox options menu item when a fragment is created. 1. I use

first()

because I only need to set the value from DataStore once, afterwards just clicking the checkbox will check/uncheck it. Does this make sense? Are there any situations where it will break? 2. Is the asynchronicity a problem? As far as I understand, you can't read from DataSore synchronously so this is the only option I can see.

override fun onCreateOptionsMenu(menu: Menu, inflater: MenuInflater) {
    
    [...]
    
    viewLifecycleOwner.lifecycleScope.launch {
        menu.findItem(R.id.action_hide_completed_tasks).isChecked =
            viewModel.preferencesFlow.first().hideCompleted
    }
}

Hello why my flow is not catched when an exception is thrown? my flow is

flow {
val resultService = service.getWhatever()
val mapStuff = resultService.mapper() <-- it throws NullPointer
emit(mapStuff)
}

Then on my viewmodel I have

usecase.execute(2)
.catch{do something}
.collect{expectedStuff}

What I'm missing? Do I have to add a 

.catch

  on my datasource so the viewmodel is aware something crashed?

I recently migrated from

PublishSubject.create<T>()

to

MutableSharedFlow<T>(replay = 0, extraBufferCapacity = 1)

, but what I can't wrap my head around is what is the use case for

MutableSharedFlow<T>()

? (both parameters set to zero by default). It doesn't really seem to do anything.

To avoid using GlobalScope in my small project, I created this application scope provided by Dagger:

@ApplicationScope
@Provides
@Singleton
fun provideApplicationScope() = CoroutineScope(SupervisorJob())

Should I also inject the dispatcher that is used with this scope, and why?

[Channels] Hi, everyone. Are channels stable? I'm getting experimental errors in InteliJ with latest coroutines dependency

implementation("org.jetbrains.kotlinx:kotlinx-coroutines-core:1.4.0-M1")

Something weird happening here, I launch a coroutine on a single thread context and the profiler shows me it creates a thread, runs the coroutine and puts that thread in WAIT, so my app is leaking threads?

I have a DialogFragment that executes some db delete operation (via its own ViewModel). Since a dialog dismisses immediately when you click a button, it makes sense to use applicationScope instead of viewModelScope, correct?

class DeleteAllCompletedViewModel @ViewModelInject constructor(
    private val taskDao: TaskDao,
    @ApplicationScope private val applicationScope: CoroutineScope
) :
    ViewModel() {

    fun deleteCompletedTasks() = applicationScope.launch {
        taskDao.deleteCompletedTasks()
    }
}

i’m a coroutine newbie and i’m stuck integrating co-routines and

CompletionStage

overall, my goal is to check whether i’ve an entity in the cache and if it’s not, go to the database the cache-checking returns a

CompletionStage

and the database repo is a suspending function how can i combine the two? the best i can do is:

val person = cache.getAsync(id).await()
return if (person == null) repository.findById(id)
else person

and obviously, it’s not the right way to do it i don’t find the correct operator help greatly appreciated 🙂

I want to insert some initial items into my todo list. Is this the right place for a custom applicationScope or should I use GlobalScope?

@Module
@InstallIn(ApplicationComponent::class)
object AppModule {
    private lateinit var database: TaskDatabase

    @Provides
    @Singleton
    fun provideDatabase(
        app: Application,
        @ApplicationScope applicationScope: CoroutineScope
    ): TaskDatabase {
        database = Room.databaseBuilder(app, TaskDatabase::class.java, "task_database")
            .fallbackToDestructiveMigration()
            .addCallback(object : RoomDatabase.Callback() {
                override fun onCreate(db: SupportSQLiteDatabase) {
                    super.onCreate(db)
                    val dao = database.taskDao()

                    applicationScope.launch {
                        dao.insert(Task("Wash the dishes"))
                        dao.insert(Task("Do the laundry"))
                        dao.insert(Task("Buy groceries"))
                        dao.insert(Task("Call mom"))
                    }
                }
            })
            .build()
        return database
    }

    @Provides
    @Singleton
    fun provideTaskDao(db: TaskDatabase) = db.taskDao()

    @ApplicationScope
    @Provides
    @Singleton
    fun provideApplicationScope() = CoroutineScope(SupervisorJob())
}

@Retention(AnnotationRetention.BINARY)
@Qualifier
annotation class ApplicationScope

How do I cancel a Flow collection and start a whole new flow?

I’m in process of migration some parts of my codebase from Rx to coroutines and I wonder if I could get the similar behavior to

private val subject = BehaviorSubject.create<Int>().startWith(Observale.just(1))

but with

SharedFlow

for example? My use case is that I want to trigger specific function(in case of coroutines

suspend

function) only once during creation of

SharedFlow

or after first collector started collecting this flow. I cannot use

onStart

or

onSubscription

because there are more than 1 collector for this flow and thus, my specific function will be triggered for every collector. So as a result a want to end up with something like

private val sharedFlow = MutableSharedFlow<Int>(replay = 1).startWith {
  // some suspending call that will be called only once
}

and also get

MutableSharedFlow

type for

sharedFlow

variable to be able to call

emit

/`tryEmit` for modification of cached value so all collectors get notified it changed. Is it possible to implement this behavior?