fix(single): predicate function receives indices starting at 0

[martinsik]

When using the predicate function in the single() operator it passes indices starting at 1 instead of 0.

Observable.from(['a', 'b', 'c', 'd', 'e'])
  .single((val, i) => {
    console.log(val + ' ' + i);
    return val === 'b';
  })
  .subscribe();

This prints to console:

a 1
b 2
c 3
d 4
e 5

Live demo: https://jsbin.com/raveji/2/edit?js,console

That's because it's first incremented (single.ts#L65) and then read again from property (single.ts#L75).

In RxJS 4 the indices start at 0 (which is probably more expected behavior). Live demo: https://jsbin.com/qihaqur/3/edit?js,console

If the user is relying on indices starting at 1 than this could be a BC. Otherwise now it's compatible with RxJS 4.

There's one more inconsistency with RxJS 4.

When there's no item matching the predicate function the RxJS 4 version sends an error notification EmptyError. Live demo: https://jsbin.com/pukikes/3/edit?js,console

Observable.range(1, 5)
  .single(val => {
    return false;
  })
  .subscribe(
    val => console.log(val),
    err => console.log('error', err)
  );

This prints to console:

error { [EmptyError: Sequence contains no elements.] message: 'Sequence contains no elements.' }

But the RxJS 5 implementation just emits undefined as next and sends error only when the single() operator received no value at all (single.ts#L87).

Live demo: https://jsbin.com/yavepan/2/edit?js,console.

The docblock for single() says:

If the source Observable emits more than one such item or no such items, notify of an IllegalArgumentException or NoSuchElementException respectively.

This describes the RxJS 4 behavior but the RxJS 5 tests cover a use-case where single() emits undefined single-spec.ts#L142. Also the two exception classes mentioned don't exist.

So I'm wondering what's the correct behavior?

[coveralls]

Coverage decreased (-0.0007%) to 97.687% when pulling b29b8e3 on martinsik:fix-single into b9f611a on ReactiveX:master.

[kwonoj]

For behavior of single, back then I was implementing unmatched one returned undefined (#322 (comment)) and discussed to update to throw Error instead. I assume I have specific path left without updating it then.

[martinsik]

@kwonoj I read the discussion and if I understand it correctly it was agreed that it should throw an error instead of undefined but it was later forgotten.

[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.