Hello all, quite simple question, I’m experimentin...

# coroutines

[antoniomarin]

Hello all, quite simple question, I’m experimenting a bit with coroutines, and I wonder what is the best approach to handle multiple endpoints inside single UseCase, I guess flow should be used?

[streetsofboston]

End points, like network requests?

[antoniomarin]

Yep, I can do the same logic if I leave objects as Single’s but I guess that not really coroutine way?

[streetsofboston]

For Singles and Maybes just use suspend fun …

[antoniomarin]

Yep but inside that one

[streetsofboston]

Not sure what you mean?

[streetsofboston]

class UseCase { 
    suspend fun getDataFromNetwork(input: Data): Response {
        val request1Result = repo.getResult1(input.field1)
        val request2Result = repo2.getResult2(input.field2)
        return combineResults(request1Result, request2Result)
    }
    ...
}

[antoniomarin]

If request1 fails then I want to try out with result2, but both should be returning same value

[streetsofboston]

Based on your example (could you post that later in this thread @antoniomarin?):

suspend fun buildUseCase(params: None) : CharSequence {
    val clientId = sessionRepository.getLoggedInClientId()
    return try {
        val content = contentsRepository.getHtmlContent(CONTENT_NAME, clientId)
        content.contentValue.toHtml()
    } catch (error: Throwable) {
        contentsRepository.getHtmlContent(CONTENT_NAME, DEFAULT_CLIENT_ID)
    }
}

[streetsofboston]

And if you need to go from a Single<T> to a Flow<T>, in case you can’t change the function/method signatures of your repos, take a look at this:
https://github.com/Kotlin/kotlinx.coroutines/tree/master/reactive/kotlinx-coroutines-rx2

[antoniomarin]

Hi @streetsofboston this is my existing code!

override fun buildUseCase(params: None): Single<CharSequence> {
    return sessionRepository.getLoggedInClientId()
        .flatMap { clientId -> contentsRepository.getHtmlContent(CONTENT_NAME, clientId) }
        .onErrorResumeNext(contentsRepository.getHtmlContent(CONTENT_NAME, DEFAULT_CLIENT_ID))
}

Thanks for helping, one question tho, how to handle error for second api which is in catch block. Does that means I would need to add other try block inside existing one?

I’m trying to figure out are there more elegant solutions 😄

[streetsofboston]

The second call to getHtmlContent is part of the onErrorResumeNext. If that second call emits an exception/throwable, the final Single returned by buildUseCase will emit that exception as well.
The same will happen in the suspend fun buildUseCase snippet as well: If the second call to getHtmlContext throws an exception, the suspend fun buildUseCase will throw an exception.
Your Rx example and the corresponding suspend example are behaving in similar ways.

[streetsofboston]

You can wrap the calls inside a runCatching block followed by a .mapCatching block, which allows you to avoid nested try-catch clauses .

[antoniomarin]

Uh haven’t heard about those two, gonna check them out btw can similar be achieved by flow, going flowOf(api1, api2) and then doing some kind of reactive flow?

Sorry if I’m making nonsenses 😄

[streetsofboston]

Using flow is overkill for async functions that just emit at most one value. In general, you’d consider:
Completable ==> suspend fun function() Unit
Maybe<T> ==> suspend fun function() : T?
Single<T> ==> suspend fun function(): T
Observable<T> or Flowable<T> ==> Flow<T>

[antoniomarin]

I ended up using something similar if someones needs it

try {
   val result = runCatching {
      repository.api(params1)
   }.getOrElse {
      repository.api(params2)
   }
   handleSuccess(result)
catch(exp: Exception) {
   handleError(exp)
}