fix(IteratorObservable): Observables from generators will now final…

[benlesh]

...ize when subscription ends

In order to model the behavior of for..of when consuming a Generator, if a break is hit in the for..of, return() is called on the generator and the generator will jump to a finally block if it has one. Observables created from generators will now have this same behavior.

    Observable.from((function* () {
      try {
        yield 1;
        yield 2;
        yield 3;
      } finally {
        console.log('finalized');
      }
    })())
    .take(2)
    .subscribe(x => console.log(x));

    // should log
    // 1
    // 2
    // finalized

fixes #1938

[coveralls]

Coverage increased (+0.002%) to 97.041% when pulling 22d286a on blesh:close_iterator into 7ee0575 on ReactiveX:master.

[jayphelps]

Anyone else reading this later, keep in mind that generators like this are basically singletons, so if one subscribers forces the generator to .return(), then every subscriber is forced into that.

const stream$ = Observable.from((function* () {
  try {
    yield 1;
    yield 2;
    yield 3;
  } finally {
    console.log('finalized');
  }
})());

stream$.take(1).subscribe();
stream$.take(3).subscribe();
// Second subscriber will only get one, not three because
// first subscriber is an scumbag

[jayphelps]

looks good, will wait a few for others to chime in since this is one of those "one person's bugfix is another person's breaking change" so shipping this at RC might be mean--but let's be honest..how many people are using generators with Rx...

[benlesh]

Related is the issue I've created in the TC39 spec. We might even want to wait for a ruling there... tc39/proposal-observable#115

[benlesh]

Given that this updates RxJS 5 to match the direction of the observable spec, I think it's the proper path. tc39/proposal-observable#115

[jayphelps]

LGTM

[lock]

This thread has been automatically locked since there has not been any recent activity after it was closed. Please open a new issue for related bugs.