obs.subscribe(next, error, complete) should bind their callbacks to undefined when present

[dead-claudia]

It seems odd and inconsistent for them to have a this that's:

1. provided by the language only
2. not undefined like pretty much everything else

Pretty much everything else in the language binds things to undefined where necessary when a function is passed as a clear non-method, so why shouldn't this?

[dead-claudia]

Oh, and BTW, this is an area where implementations differ:

• RxJS' this is a private observer that shouldn't have been exposed. Edit: link
• zen-observable and fate-observable both follow the spec correctly by ensuring this is that newly created object.

[benjamingr]

Well, honestly I always aw observables as:

class Observable {
  constructor(subscribe) { this.subscribe = subscribe; }
}

Which is a "minimal observable implementation" (just a function reference I can call to). In that scenario when calling .subscribe on the observable - I would expect the this value to be the observable I am subscribing to.

That said, I don't recall ever using it in any way ever.

[dead-claudia]

@benjamingr Neither do I, and personally, I wouldn't miss it if that 3-argument prototype disappeared from the spec altogether. It's not like very many have it - most implementations that don't try to implement the spec don't even bother to implement the 3-argument variant, just the one.

[dead-claudia]

I think the only times I've used it were in toy examples. The problem with it is that it's not obvious from first sight which one's next, which one's error, and which one's complete. It's something you could give a pass for with promises since there's only two (it's hard to miss a divider between two functions), but I still find myself occasionally having to do a double-take to remember it.

[appsforartists]

Considering the observer variant is effectively named args, I don't see a need at all for the 3 arg variant:

$.subscribe({
  next(value) {...},
  complete() {...},
  error(error) {...},
})

is a more readable/easier to understand alternative that's already supported.