Bug 216331 – Port BaseAudioContext to the HTML event loop

• Status: RESOLVED
• Resolution: FIXED
• Component: Web Audio

[Chris Dumez]

Port BaseAudioContext to the HTML event loop.

[Chris Dumez]

Created attachment 408377 [details]
Patch

[Chris Dumez]

Created attachment 408389 [details]
Patch

[Chris Dumez]

Created attachment 408392 [details]
Patch

[Chris Dumez]

Created attachment 408394 [details]
Patch

[Chris Dumez]

Created attachment 408449 [details]
Patch

[Eric Carlson]

Comment on attachment 408449 [details]
Patch

View in context: https://bugs.webkit.org/attachment.cgi?id=408449&action=review

> LayoutTests/webaudio/audiocontext-state-interrupted.html:74
> +    setTimeout(function() {
> +        debug('');
> +        debug('Test 2: resume() while interrupted will cause interruption to end.')
>  
> -    context.onstatechange = secondInterruptionStarted;
> -    if (window.internals)
> -        evalAndLog('internals.beginMediaSessionInterruption("System")');
> +        context.onstatechange = secondInterruptionStarted;
> +        if (window.internals)
> +            evalAndLog('internals.beginMediaSessionInterruption("System")');
> +    }, 0);

It would be good to have a LayoutTests ChangeLog explaining why these changes are necessary.

[Chris Dumez]

(In reply to Eric Carlson from comment #6)
> Comment on attachment 408449 [details]
> Patch
> 
> View in context:
> https://bugs.webkit.org/attachment.cgi?id=408449&action=review
> 
> > LayoutTests/webaudio/audiocontext-state-interrupted.html:74
> > +    setTimeout(function() {
> > +        debug('');
> > +        debug('Test 2: resume() while interrupted will cause interruption to end.')
> >  
> > -    context.onstatechange = secondInterruptionStarted;
> > -    if (window.internals)
> > -        evalAndLog('internals.beginMediaSessionInterruption("System")');
> > +        context.onstatechange = secondInterruptionStarted;
> > +        if (window.internals)
> > +            evalAndLog('internals.beginMediaSessionInterruption("System")');
> > +    }, 0);
> 
> It would be good to have a LayoutTests ChangeLog explaining why these
> changes are necessary.

Yes. Basically the issue is we sometimes move to on test to another based a Promise resolution (e.g. context.resume() / context.suspend() return promises). The promise gets resolved *BEFORE* the statechange event gets fired, as per spec. The setTimeout(0) makes sure any pending statechange event has been fired before we move on to the next text (because the next test may rely on the statechange event and we don't want it to receive the event from the previous test).

[Chris Dumez]

Created attachment 408450 [details]
Patch

[EWS]

Committed r266835: <https://trac.webkit.org/changeset/266835>

All reviewed patches have been landed. Closing bug and clearing flags on attachment 408450 [details].

[Radar WebKit Bug Importer]

<rdar://problem/68650688>