RxJava `Completable.andThen` is not executing serially? - java

I have a usecase where I initiallize some global variables in a Completable , and in the next step in the chain (using andThen operator) I make use of those variables.
Following sample explains my usecase in detail
Say you have a class User
class User {
String name;
}
and I have an Observable like this ,
private User mUser; // this is a global variable
public Observable<String> stringObservable() {
return Completable.fromAction(() -> {
mUser = new User();
mUser.name = "Name";
}).andThen(Observable.just(mUser.name));
}
First I'm doing some initiallizations in my Completable.fromAction and I expect andThen operator to start only after completing the Completable.fromAction.
Which means I expect mUser to be initallized when the andThen operator starts.
Following is my subscription to this observable
stringObservable()
.subscribe(s -> Log.d(TAG, "success: " + s),
throwable -> Log.e(TAG, "error: " + throwable.getMessage()));
But when I run this code , I get an error
Attempt to read from field 'java.lang.String User.name' on a null object reference
which means mUser is null , andThen started before executing the code in Completable.fromAction. Whats happening here?
According to documentation of andThen
Returns an Observable which will subscribe to this Completable and once that is completed then will subscribe to the {#code next} ObservableSource. An error event from this Completable will be propagated to the downstream subscriber and will result in skipping the subscription of the Observable.

The issue is not with andThen but with the statement Observable.just(mUser.name) inside andThen . The just operator will try to create the observable immediately though it will emit only after Completable.fromAction.
Problem here is , while trying to create the Observable using just , the mUser is null.
Solution : You need to defer the creation of the String Observable till a subscription happens , till the upstream of andThen starts emission.
Instead of andThen(Observable.just(mUser.name));
use
andThen(Observable.defer(() -> Observable.just(mUser.name)));
Or
andThen(Observable.fromCallable(() -> mUser.name));

I don't think #Sarath Kn's answer is 100% correct. Yes just will create observable as soon as it's called, but andThen is still calling just at an unexpected time.
We can compare andThen with flatMap to get some better understanding. Here is a fully runnable test:
package com.example;
import org.junit.Test;
import io.reactivex.Completable;
import io.reactivex.Observable;
import io.reactivex.observers.TestObserver;
import io.reactivex.schedulers.Schedulers;
public class ExampleTest {
#Test
public void createsIntermediateObservable_AfterSubscribing() {
Observable<String> coldObservable = getObservableSource()
.flatMap(integer -> getIntermediateObservable())
.subscribeOn(Schedulers.trampoline())
.observeOn(Schedulers.trampoline());
System.out.println("Cold obs created... subscribing");
TestObserver<String> testObserver = coldObservable.test();
testObserver.awaitTerminalEvent();
/*
Resulting logs:
Creating observable source
Cold obs created... subscribing
Emitting 1,2,3
Creating intermediate observable
Creating intermediate observable
Creating intermediate observable
Emitting complete notification
IMPORTANT: see that intermediate observables are created AFTER subscribing
*/
}
#Test
public void createsIntermediateObservable_BeforeSubscribing() {
Observable<String> coldObservable = getCompletableSource()
.andThen(getIntermediateObservable())
.subscribeOn(Schedulers.trampoline())
.observeOn(Schedulers.trampoline());
System.out.println("Cold obs created... subscribing");
TestObserver<String> testObserver = coldObservable.test();
testObserver.awaitTerminalEvent();
/*
Resulting logs:
Creating completable source
Creating intermediate observable
Cold obs created... subscribing
Emitting complete notification
IMPORTANT: see that intermediate observable is created BEFORE subscribing =(
*/
}
private Observable<Integer> getObservableSource() {
System.out.println("Creating observable source");
return Observable.create(emitter -> {
System.out.println("Emitting 1,2,3");
emitter.onNext(1);
emitter.onNext(2);
emitter.onNext(3);
System.out.println("Emitting complete notification");
emitter.onComplete();
});
}
private Observable<String> getIntermediateObservable() {
System.out.println("Creating intermediate observable");
return Observable.just("A");
}
private Completable getCompletableSource() {
System.out.println("Creating completable source");
return Completable.create(emitter -> {
System.out.println("Emitting complete notification");
emitter.onComplete();
});
}
}
You can see that when we use flatmap, the just is called after subscribing, which makes sense. If the intermediate observable depended on the items emitted to the flatmap then of course the system can't create the intermediate observable before subscription. It would not yet have any values. You can imagine this wouldn't work if flatmap called just before subscribing:
.flatMap(integer -> getIntermediateObservable(integer))
What is weird is that andThen is able to create it's inner observable (i.e. call just) before subscribing. It makes sense that it can do this. The only thing andThen is going to receive is a complete notification, so there is no reason NOT to create the intermediate observable early. The only problem is that it's not the expected behavior.
#Sarath Kn's solution is correct, but for the wrong reason. If we use defer we can see things working as expected:
#Test
public void usingDefer_CreatesIntermediateObservable_AfterSubscribing() {
Observable<String> coldObservable = getCompletableSource()
.andThen(Observable.defer(this::getIntermediateObservable))
.subscribeOn(Schedulers.trampoline())
.observeOn(Schedulers.trampoline());
System.out.println("Cold obs created... subscribing");
TestObserver<String> testObserver = coldObservable.test();
testObserver.awaitTerminalEvent();
/*
Resulting logs:
Creating completable source
Cold obs created... subscribing
Emitting complete notification
Creating intermediate observable
IMPORTANT: see that intermediate observable is created AFTER subscribing =) YEAY!!
*/
}

Related

Fire and forget Action on doOnNext in Project Reactor

I have a Flux stream. For each element processed I want to have an action triggered which is an asynchronous/non-blocking one. For example, a method returning back a Mono from a db update.
I want this action to be done on the doOnNext block.
I don't want to affect the Flux, the processing and the back pressure implemented there.
Supposing Mono method to be called is
Mono<Integer> dbUpdate();
should my Flux be like this?
public Flux<Data> processData(PollRequest request)
{
return searchService.search(request)
.doOnNext(data -> dbUpdate(data));
}
Or should be as mentioned on a stack overflow example.
public Flux<Data> processData(PollRequest request)
{
return searchService.search(request)
.doOnNext(data -> dbUpdate(data).subscribe());
}
Won't the above cause blocking issues inside doOnNext?
Also which is the most appropriate scheduler to use for this type of action?
dbUpdate() will be ignored if you do not subscribe to it. The following snippet doesn't print anything because Mono.just("db update") doesn't get subscribed.
Mono<String> dbUpdate() {
return Mono.just("db update")
.doOnNext(System.out::println);
}
public Flux<String> processData() {
return Flux.just("item 1", "item 2")
.doOnNext(data -> dbUpdate());
}
Note that .subscribe() doesn't block your thread, it kicks off the work and returns immediately.

How to capture the reactive stream cancel signal?

I want to do something finally after stream terminates for any reason including cancellation, and I
found the doFinally method, but it dose not work when cancellation, because https://github.com/reactor/reactor-core/issues/1090#issuecomment-367633241 show :
Cancellation travels only upstream
So, how to capture the cancel signal?
There is my code:
public Mono<Void> myFunction() {
return Mono.just("hello")
.flatMap(s -> foo(s))
.doFinally(signalType -> {
// do something finally, but the doFinally won't be called
System.out.println(signalType);
});
}
// some other library's function that I cant not modify any way
public Mono<Void> foo(String s) {
// return a reactive stream, and will cancel it after it be subscribed, like:
return Mono.just(s)
.doOnSubscribe(subscription -> subscription.cancel())
.then();
}
You can't in that particular arrangement, because the foo() method/library seems to manage the subscription (the cancellation) itself, instead of leaving that responsibility to the consumer. Managing the subscription like that is thus not necessarily a good thing.

Equivalent of VertX CompositeFuture in RxJava

The VertX example for when you need to query multiple asynchronous resources and use them all in a single operation is:
Future<HttpServer> httpServerFuture = Future.future();
httpServer.listen(httpServerFuture.completer());
Future<NetServer> netServerFuture = Future.future();
netServer.listen(netServerFuture.completer());
CompositeFuture.all(httpServerFuture, netServerFuture).setHandler(ar -> {
if (ar.succeeded()) {
// All servers started
} else {
// At least one server failed
}
});
We need to query two different databases and then use the results in business logic, but the flow is equivalent.
What's the VertX/RxJava equivalent?
Currently people are doing this by nesting a new .flatMap() call every time they need a new variable. I'm left feeling there must be a better way...
We don't actually need the queries to be concurrent but we need to cache both results and pass them to the business logic at the same time some how.
there are many ways to do this, but i've tried to pick an approach that tacks closely to your sample:
#Override
public void start(Future<Void> startFuture) throws Exception {
final HttpServer httpServer = vertx.createHttpServer();
final Completable initializeHttpServer = httpServer.rxListen().toCompletable();
final NetServer netServer = vertx.createNetServer();
final Completable initializeNetServer = netServer.rxListen().toCompletable();
initializeHttpServer.andThen(initializeNetServer)
.subscribe(
() -> { /* All servers started */ },
error -> { /* At least one server failed */ }
);
}
the rxListen() invocations are converted into Completable instances, which are then run serially upon subscription.
the subscriber's onComplete callback will be invoked when both servers are done binding to their respective ports, or...
the onError callback will be invoked if an exception occurs
(also, fwiw, "nesting" flatMap operations for something as trivial as this shouldn't be necessary. "chaining" such operations, however, would be idiomatic usage).
hope that helps!
--UPDATE--
having read the question more carefully, i now see that you were actually asking about how to handle the results of two discrete asynchronous operations.
an alternative to flatMap'ing your way to combining the results would be to use the zip operator, like so:
#Override
public void start(Future<Void> startFuture) throws Exception {
final Single<String> dbQuery1 = Single.fromCallable(() -> { return "db-query-result-1"; });
final Single<String> dbQuery2 = Single.fromCallable(() -> { return "db-query-result-2"; });
Single.zip(dbQuery1, dbQuery2, (result1, result2) -> {
// handle the results from both db queries
// (with Pair being a standard tuple-like class)
return new Pair(result1, result2);
})
.subscribe(
pair -> {
// handle the results
},
error -> {
// something went wrong
}
);
}
per the docs, zip allows you to specify a series of reactive types (Single, Observable, etc) along with a function to transform all the results at once, with the central idea being that it will not emit anything until all the sources have emitted once (or more, depending on the reactive type).

How to test non-RxJava observables or async code in general?

I'm playing around with implementing my own observables or porting them from other languages for fun and profit.
The problem I've run into is that there's very little info on how to properly test observables or async code in general.
Consider the following test code:
// Create a stream of values emitted every 100 milliseconds
// `interval` uses Timer internally
final Stream<Number> stream =
Streams.interval(100).map(number -> number.intValue() * 10);
ArrayList<Number> expected = new ArrayList<>();
expected.add(0);
expected.add(10);
expected.add(20);
IObserver<Number> observer = new IObserver<Number>() {
public void next(Number x) {
assertEquals(x, expected.get(0));
expected.remove(0);
if(expected.size() == 0) {
stream.unsubscribe(this);
}
}
public void error(Exception e) {}
public void complete() {}
};
stream.subscribe(observer);
As soon as the stream is subscribed to, it emits the first value. onNext is called... And then the test exits successfully.
In JavaScript most test frameworks nowadays provide an optional Promise to the test case that you can call asynchronously on success/failure. Is anything similar available for Java?
Since the execution is asyncronious, you have to wait until is finish. You can just wait for some time in an old fashion way
your_code
wait(1000)
check results.
Or if you use Observables you can use TestSubscriber
In this example you can see how having an async operation we wait until the observer consume all items.
#Test
public void testObservableAsync() throws InterruptedException {
Subscription subscription = Observable.from(numbers)
.doOnNext(increaseTotalItemsEmitted())
.subscribeOn(Schedulers.newThread())
.subscribe(number -> System.out.println("Items emitted:" + total));
System.out.println("I finish before the observable finish. Items emitted:" + total);
new TestSubscriber((Observer) subscription)
.awaitTerminalEvent(100, TimeUnit.MILLISECONDS);
}
You can see more Asynchronous examples here https://github.com/politrons/reactive/blob/master/src/test/java/rx/observables/scheduler/ObservableAsynchronous.java

correct way to join an rxjava subscription with in flight observable

I have a method getData() that returns an Observable.
public Observable<Data> getData() {
Observable<Data> observable = mApi.networkCall();
return observable;
}
There are many clients that call this method, and if the observable has not completed yet, I'd like them to just subscribe to the in flight observable. If the observable has completed, I need to restart the call and get a new observable. What's the best way to detect the observable is complete and i should call the network call again?
Use .share(). The first subscription starts the observable and subsequent subscribers will receive the same emission(s). Once the stream terminates all current subscribers will be invalid and the next subscription will start the observable again.
public Observable<Data> getData() {
return mApi.networkCall().share();
}
Note that I'm only expecting getData() to be called once and the returned observable reused. If that wasn't the case we'd make sure that getData() returned a singleton.

Categories