Synchronous emit in flush method of debounce produces incorrect result

[betalb]

If you will try to do synchronous emit in debounced observable incorrect sequence of events will be produced

i.e.

const a = Kefir.pool();
const b = a.debounce(100).spy();
a.plug(b.filter((val) => val === 1).map(() => 2));
a.plug(Kefir.constant(1));
a.observe();

Expected output

100ms [pool.debounce] <value> 1
200ms [pool.debounce] <value> 2

Actual output

100ms [pool.debounce] <value> 1
200ms [pool.debounce] <value> null

[mAAdhaTTah]

Without having investigated, the issue is probably less about synchronous emit and more about the circular dependency between a & b.

[betalb]

yes, there is a cycle, but it is not infinite and is actually done to illustrate an issue

Here is another more dirty example that produces the same result

let e;
Kefir.stream((emitter) => { e = emitter; })
  .debounce(100)
  .withHandler((emitter, event) => {
    if (event.type === 'value' && event.value === 1) {
      e.emit(2);
    }
    emitter.emitEvent(event)
  }).log();
e.emit(1);

Actually I created PR that should highlight where the issues seems to be. And there are similar problems in buffer-by, buffer-while-by (I just wanted to create cleaner test case before creating separate ticket)

[rpominov]

This is documented behaviour:

Although documentation might be more clear probably. My idea was back then is that synchronous events is not a use case I wanted to support well in the library in general.

[betalb]

@rpominov I'm not following documentation quote, for me it sound like a different case, which I never faced 🙂

This issue is not with the fact that some value is emitted without being debounced, but completely wrong value (null) being emitted

Also I've provided 2nd example that doesn't use properties and here is 1st example without using properties either

const a = Kefir.pool();
const b = a.debounce(100).spy();
a.plug(b.filter((val) => val === 1).map(() => 2));
a.plug(Kefir.stream(emitter => { emitter.emit(1) }));
a.observe();

[mAAdhaTTah]

@rpominov The issue isn't that it's emitted synchronously (it actually isn't, according to the issue description), but that the second event is dropped.

[rpominov]

Ah, right, sorry

[bpinto]

Hey guys, @betalb opened a PR fixing this, do you think we could merge their PR?

[mAAdhaTTah]

@bpinto Done. Let me get a release out this weekend.