subscriber methods with scheduler create ambiguous overloads for Groovy and Java 8

[ldaley]

Hi,

In 0.18 there are the following methods:

public final Subscription subscribe(final Action1<? super T> onNext, final Action1<Throwable> onError)
public final Subscription subscribe(final Action1<? super T> onNext, Scheduler scheduler)

The “problem” here is that Action1 and Scheduler are both SAM types. This makes closure coercions and lambda expressions ambiguous and requires extra syntax to specify the coercion target type.

Could the Scheduler accepting methods be renamed?

There are other similar methods of course.

[benjchristensen]

@headinthebox and @akarnokd What is the reason for having the Scheduler overloads on the subscribe methods? I honestly can not remember why that happened. Seems like that should just be using subscribeOn and does not need these overloads.

This is not something we want to leave as is because it breaks Groovy and Clojure.

[ldaley]

It also makes it slightly less convenient to use a Java 8 lambda expressions because you need to add the target type to the expression.

[headinthebox]

Never understood why, this it not in .net ... I would junk that overload. Not even sure what it is supposed to do.

[akarnokd]

Looks like a convenience shortcut and we should avoid ambiguity problems with Java 8 while RxJava is in a "allowed" API flux. I vote for removal.

[benjchristensen]

So do I, let's remove them.

[ldaley]

Would it be worth some tests that reflectively check the public API for any methods that are overloaded with different SAM types (i.e. this problem) to catch this before leaking out again?

[benjchristensen]

Yes, a test like that would be good. You interesting in doing it?

[ldaley]

Sure, I'll do it.

I'll do something for the methods on Observable, it could then later be adapted to other applicable types.

[benjchristensen]

Thanks, and I'll delete the methods shortly.