I'm trying to understand the purpose of this library by Jake Warthon:
https://github.com/JakeWharton/RxRelay
Basically: A Subject except without the ability to call onComplete or
onError. Subjects are stateful in a damaging way: when they receive an
onComplete or onError they no longer become usable for moving data.
I get idea, it's a valid use case, but the above seems easy to achieve just using the existing subjects.
1. Don't forward errors/completions events to the subject:
`observable.subscribe({ subject.onNext(it) }, { log error / throw exception },{ ... })`
2. Don't expose the subject, make your method signature return an observable instead.
fun(): Observable<> { return subject }
I'm obviously missing something here and I'm very curios on what it is!
class MyPublishRelay<I> : Consumer<I> {
private val subject: Subject<I> = PublishSubject.create<I>()
override fun accept(intent: I) = subject.onNext(intent)
fun subscribe(): Disposable = subject.subscribe()
fun subscribe(c: Consumer<in I>): Disposable = subject.subscribe(c)
//.. OTHER SUBSCRIBE OVERLOADS
}
subscribe has overloads and, usually, people get used to the subscribe(Consumer) overload. Then they use subjects and suddenly onComplete is also invoked. RxRelay saves the user from themselves who don't think about the difference between subscribe(Consumer) and subscribe(Observer).
Don't forward errors/completions events to the subject:
Indeed, but based on our experience with beginners, they often don't think about this or even know about the available methods to consider.
Don't expose the subject, make your method signature return an observable instead.
If you need a way to send items into the subject, this doesn't work. The purpose is to use the subject to perform item multicasting, sometimes from another Observable. If you are in full control of the emissions through the Subject, you should have the decency of not calling onComplete and not letting anything else do it either.
Subjects have far more overhead because they have to track and handle
terminal event states. Relays are stateless aside from subscription
management.
- Jake Wharton
(This is from the issue OP opened on GitHub and felt it was a more a correct answer and wanted to "relay" it here for others to see. https://github.com/JakeWharton/RxRelay/issues/30)
In addition to #akarnokd answer:
In some cases you can't control the flow of data inside the Observable, an example of this is when observing data changes from a database table using Room Database.
If you use Subjects, executing subjects.getValue will always throw error about null safety. So you have to put "? or !!" everywhere in your code even though you know that it will be not nullable.
public T getValue() {
Object o = value.get();
if (NotificationLite.isComplete(o) || NotificationLite.isError(o)) {
return null;
}
return NotificationLite.getValue(o);
}
Related
I have some code in a class that takes FileSystemWatcher events and flattens them into an event in my domain:
(Please note, the *AsObservable methods are extensions from elsewhere in my project, they do what they say 🙂.)
watcher = new FileSystemWatcher(ConfigurationFilePath);
ChangeObservable = Observable
.Merge(
watcher.ChangedAsObservable().Select((args) =>
{
return new ConfigurationChangedArgs
{
Type = ConfigurationChangeType.Edited,
};
}),
watcher.DeletedAsObservable().Select((args) =>
{
return new ConfigurationChangedArgs
{
Type = ConfigurationChangeType.Deleted,
};
}),
watcher.RenamedAsObservable().Select((args) =>
{
return new ConfigurationChangedArgs
{
Type = ConfigurationChangeType.Renamed,
};
})
);
ChangeObservable.Subscribe((args) =>
{
Changed.Invoke(this, args);
});
Something that I'm trying to wrap my head around as I'm learning are best practices around naming, ownership and cleanup of the IObservable and IDisposable returned by code like this.
So, some specific questions:
Is it okay to leak IObservables from a class that creates them? For example, is the property I'm assigning this chain to okay to be public?
Does the property name ChangeObservable align with what most people would consider best practice when using the .net reactive extensions?
Do I need to call Dispose on any of my subscriptions to this chain, or is it safe enough to leave everything up to garbage collection when the containing class goes out of scope? Keep in mind, I'm observing events from watcher, so there's some shared lifecycle there.
Is it okay to take an observable and wire them into an event on my own class (Changed in the example above), or is the idea to stay out of the native .net event system and leak my IObservable?
Other tips and advice always appreciated! 😀
Is it okay to leak IObservables from a class that creates them? For
example, is the property I'm assigning this chain to okay to be
public?
Yes.
Does the property name ChangeObservable align with what most
people would consider best practice when using the .net reactive
extensions?
Subjective question. Maybe FileChanges? The fact that it's an observable is clear from the type.
Do I need to call Dispose on any of my subscriptions to
this chain, or is it safe enough to leave everything up to garbage
collection when the containing class goes out of scope?
The ChangeObservable.Subscribe at the end could live forever, preventing the object from being garbage collected if the event is subscribed to, though that could also be your intention. Operator subscriptions are generally fine. I can't see the code for your ChangedAsObservable like functions. If they don't include a Subscribe or an event subscription, they're probably fine as well.
Keep in mind,
I'm observing events from watcher, so there's some shared lifecycle
there.
Since FileWatcher implements IDisposable, you should probably use Observable.Using around it so you can combine the lifecycles.
Is it okay to take an observable and wire them into an event on
my own class (Changed in the example above), or is the idea to stay
out of the native .net event system and leak my IObservable?
I would prefer to stay in Rx. The problem with event subscriptions is that they generally live forever. You lose the ability to control subscription lifecycle. They're also feel so much more primitive. But again, that's a bit subjective.
Implementations of rx provide BehaviorSubject<T> and Variable<T> as mechanisms for modeling properties that change over time (a useful replacement for C# INotifyPropertyChanged).
Generally these are exposed as Observable<T> but it would be more useful to expose properties as something like:
class ObservableValue<T> : Observable<T>{
var currentValue:T { get }
}
This can be created along these lines in swift:
class ObservableValue<Element> : ObservableType {
typealias E = Element
private let subject:BehaviorSubject<E>
var currentValue:E {
get {
return try! subject.value()
}
}
init(subject:BehaviorSubject<E>) {
self.subject = subject
}
func subscribe<O: ObserverType where O.E == E>(observer: O) -> Disposable {
return self.subject.subscribe(observer)
}
}
Does this already exist? and if not is it because it's against the aims of Rx?
The only way around it is to expose a separate currentValue or write consumers that assume the concrete implementation behind the exposed Observable is a BehaviourSubject or somewhere in the chain a replay() has occured e.g. the following snippet doesn't make it explicit that as soon as I subscribe I will get a value:
class MyViewModel {
// 'I will notify you of changes perhaps including my current value'
myProperty:Observable<String>
}
so code has to be written as if its 'asynchronous' with an underlying assumption it will act in an almost synchronous manner rather than:
class MyViewModel {
// 'I have a current value and will notify you of changes going forward'
myProperty:ObservableValue<String>
}
Having thought it over and discussed it a bit more presumably the reason it doesn't (and perhaps shouldn't exist) is that it's an introduction of imperatively accessed state.
Other mechanisms of maintaining state (such as scan) do so within the confines of chained observables rather than as 'dead-end' direct calls such as 'give me the value right now'.
Perhaps it would have it's place in a hybrid reactive/imperative approach but it may just hinder full embracement of the reactive style.
It's analogous to using promises or tasks in half of the code then reverting to synchronous blocking code in other parts.
In most cases what people do is create a standard view model that exposes properties via INotifyPropertyChanged. This allows UI elements to bind to them and receive property change events and keep the UI in sync.
Then if you want an IObservable for said property you take advantage of standard Rx operators that turn events into IObservable. You can google this to find lots of different implementations. You would generally create and consume these observables from something that is observing the view model rather than expose them on the view model directly.
So I'm pretty new to both Scala and RX. The guy who knew the most, and who actually wrote this code, just left, and I'm not sure what's going on. This construct is all over his code and I'm not really clear what its doing:
def foo(List[Long]) : Observable[Unit] =
Observable {
subscriber => {
do some stuff
subscriber.onNext()
subscriber.onCompleted()
}
I mostly get do some stuff, and the calls to subscriber. What I don't get is, where does subscriber come from? Does subscriber => { instantiate the subscriber? What does Observable { subscriber => { ... } } do/mean?
If you take a look at the Observable companion object documentation, you will see an apply method that takes a function of type (Subscriber[T]) ⇒ Unit. So, when you call Observable{withSomeLambda}, then this is the same as calling Observable.apply{withSomeLambda}
And, if you go all the way to the source code you will see that this is really returning
toScalaObservable(rx.Observable.create(f))
where f is the lambda that you passed in.
So, subscriber is just the parameter of the lambda. It is passed in by the caller of that function.
This code is creating a new Observable as described here.
Basically when a downstream component subscribes to this stream, this callback is called. In the callback we determine when we, as a data source, will call onNext(v: T) which is how we pass each element we are generating to them, and when we will call onCompleted() which is how we tell the subscriber that we are done sending data.
Once you have created an Observable you can start calling Observable operators, which will either result in another, compound Observable, or will result in a terminating condition which will end the process, and generally result in a final result for the flow (often a collection or aggregate value).
You don't use the List in your question, but normally if you wanted to make a reactive stream out of a list you would call Observable.from().
PS: I think this is RxJava code.
I'm using GWTP, adding a Contract layer to abstract the knowledge between Presenter and View, and I'm pretty satisfied of the result with GWTP.
I'm testing my presenters with Mockito.
But as time passed, I found it was hard to maintain a clean presenter with its tests.
There are some refactoring stuff I did to improve that, but I was still not satisfied.
I found the following to be the heart of the matter :
My presenters need often asynchronous call, or generally call to objects method with a callback to continue my presenter flow (they are usually nested).
For example :
this.populationManager.populate(new PopulationCallback()
{
public void onPopulate()
{
doSomeStufWithTheView(populationManager.get());
}
});
In my tests, I ended to verify the population() call of the mocked PopulationManager object. Then to create another test on the doSomeStufWithTheView() method.
But I discovered rather quickly that it was bad design : any change or refactoring ended to broke a lot of my tests, and forced me to create from start others, even though the presenter functionality did not change !
Plus I didn't test if the callback was effectively what I wanted.
So I tried to use mockito doAnswer method to do not break my presenter testing flow :
doAnswer(new Answer(){
public Object answer(InvocationOnMock invocation) throws Throwable
{
Object[] args = invocation.getArguments();
((PopulationCallback)args[0]).onPopulate();
return null;
}
}).when(this.populationManager).populate(any(PopulationCallback.class));
I factored the code for it to be less verbose (and internally less dependant to the arg position) :
doAnswer(new PopulationCallbackAnswer())
.when(this.populationManager).populate(any(PopulationCallback.class));
So while mocking the populationManager, I could still test the flow of my presenter, basically like that :
#Test
public void testSomeStuffAppends()
{
// Given
doAnswer(new PopulationCallbackAnswer())
.when(this.populationManager).populate(any(PopulationCallback.class));
// When
this.myPresenter.onReset();
// Then
verify(populationManager).populate(any(PopulationCallback.class)); // That was before
verify(this.myView).displaySomething(); // Now I can do that.
}
I am wondering if it is a good use of the doAnswer method, or if it is a code smell, and a better design can be used ?
Usually, my presenters tend to just use others object (like some Mediator Pattern) and interact with the view. I have some presenter with several hundred (~400) lines of code.
Again, is it a proof of bad design, or is it normal for a presenter to be verbose (because its using others objects) ?
Does anyone heard of some project which uses GWTP and tests its presenter cleanly ?
I hope I explained in a comprehensive way.
Thank you in advance.
PS : I'm pretty new to Stack Overflow, plus my English is still lacking, if my question needs something to be improved, please tell me.
You could use ArgumentCaptor:
Check out this blog post fore more details.
If I understood correctly you are asking about design/architecture.
This is shouldn't be counted as answer, it's just my thoughts.
If I have followed code:
public void loadEmoticonPacks() {
executor.execute(new Runnable() {
public void run() {
pack = loadFromServer();
savePackForUsageAfter();
}
});
}
I usually don't count on executor and just check that methods does concrete job by loading and saving. So the executor here is just instrument to prevent long operations in the UI thread.
If I have something like:
accountManager.setListener(this);
....
public void onAccountEvent(AccountEvent event) {
....
}
I will check first that we subscribed for events (and unsubscribed on some destroying) as well I would check that onAccountEvent does expected scenarios.
UPD1. Probably, in example 1, better would be extract method loadFromServerAndSave and check that it's not executed on UI thread as well check that it does everything as expected.
UPD2. It's better to use framework like Guava Bus for events processing.
We are using this doAnswer pattern in our presenter tests as well and usually it works just fine. One caveat though: If you test it like this you are effectively removing the asynchronous nature of the call, that is the callback is executed immediately after the server call is initiated.
This can lead to undiscovered race conditions. To check for those, you could make this a two-step process: when calling the server,the answer method only saves the callback. Then, when it is appropriate in your test, you call sometinh like flush() or onSuccess() on your answer (I would suggest making a utility class for this that can be reused in other circumstances), so that you can control when the callback for the result is really called.
I'm connecting to an object that asyncronously loads a collection of objects into an IEnumerable. At the time I connect, the IEnumerable may have items already in it's collection, and may add items during the lifetime of the application that I need to be notified of as they occur. As an example, it could be a bank account containing a list of bank transactions.
The challenge is this. I want to combine the processing of the initial values in the IEnumerable with any new additions. They are currently two processes. I would like to eliminate the use of NotifyCollectionChanged entirely.
I can modify the backend holding the IEnumerable. It does not need to remain as an IEnumerable if a solution to this question exists otherwise.
I would suggest that the object should not expose a IEnumerable as that is for "cold observable values", where in your case you need something which can get additional items in future also.
The best way to model this would be to use ReplaySubject<T> instead of IEnumerable. Below is an example that demonstrate the situation similar of yours:
//Function to generate the subject with future values
public static ReplaySubject<int> GetSubject()
{
var r = new ReplaySubject<int>();
r.OnNext(1); r.OnNext(2); r.OnNext(3);
//Task to generate future values
Task.Factory.StartNew(() =>
{
while (true)
{
Thread.Sleep(3000);
r.OnNext(DateTime.Now.Second);
}
});
return r;
}
Consuming code:
var sub = GetSubject();
sub.Subscribe(Console.WriteLine);
Every time anyone subscribes to sub they will get all the values that have been published in the subject till now and as well as new values that this subject generates in future
You can use Defer/Replay Operator