Re: [Agenda] WG/CG Telecon Thurs May 06, 9:00 am PDT

Attendees:  Michel Buffa, Chris Lilley, Raymond Toy, Ruth John, Paul
Adenot, Christoph Guttandin

Minutes:


   - V1
      - CR and Rec Status
         - PR
            - ChrisL: Still waiting for Mozilla?
            - Paul: Working on it
            - Raymond: Still need one from me as co-chair?
            - ChrisL: That would be nice, but not necessary.
            - ChrisL: PR got published today.
         - Living standard
         <https://github.com/WebAudio/web-audio-api/issues/2320>
            - Raymond: I think everyone thinks this is a good idea.
            - Raymond: How do we update the spec then?
            - ChrisL: Flag inline additions for review.  Stays as part of
            recommendation.
            - Raymond: How is it flagged?
            - ChrisL: Update current proposed spec with new stuff.  Get
            reviews and such and then create the recommendation spec
and publish it.
            - ChrisL: WebAudio will be one of the first groups to do this.
            - Raymond: How do we mark the new stuff?
            - ChrisL: Bikeshed has some way to do this so you can skim
            through to find the new stuff.
            - Raymond: I'm happy with this.
         - New Issues
         - ECMA262 Integrity Level
         <https://github.com/WebAudio/web-audio-api/issues/2340>
            - [Fix for broken link from Chris Lilley]
         - ECMA262 Completion links
         <https://github.com/WebAudio/web-audio-api/issues/2339>
            - [Fix for broken link from Chris Lilley]
            - WebIDL no longer defines "present"
         <https://github.com/WebAudio/web-audio-api/issues/2338>
            - [Fix for broken link from Chris Lilley]
            - Raymond: This looks a bit more complex
            - Chris Lilley: I tried to find where it changed but didn't
            find it.  This is my best guess on what it should be.
            - Raymond: I think we need to review this a bit more.
         - outputLatency normative, non-optional
         <https://github.com/WebAudio/web-audio-api/issues/2330>
            - Paul: Can't do A/V sync without this, right?
            - ChrisG: Safari supports this now.
            - Paul: MediaElement handles this for you.
            - Paul: Another place it's useful is AEC. Zoom  wants this to
            do their own stuff.  You need end-to-end latency to speed up the
            correlation for AEC.
            - Raymond: Does input have a latency value?
            - Paul/Chris: Yes, there's a way to get this.
            - Raymond: Conclusion: It will not be optional; it's really
            needed.
            - Current issues
         - decodeAudioData and corrupted files
         <https://github.com/WebAudio/web-audio-api/issues/2321>
            - Raymond: Issue closed, but should we add a note?
            - Paul: Yes, we can add something, but can't change behavior
            - Michel: Can't we throw an error?
            - Paul: We can't change behavior now; it's been there for 10+
            years.
            - Paul: Should it be reopened?
            - [Paul/Raymond think about how to say this, but can't come up
            with an answer quickly.]
            - Raymond: I can do  that; we'll come up with some appropriate
            text.
         - V2
      - 16 Priority-1 issues
      <https://github.com/WebAudio/web-audio-api-v2/issues?q=is%3Aopen+is%3Aissue+label%3Apriority-1>
         - Render capacity
         <https://github.com/WebAudio/web-audio-api-v2/issues/40>
            - Raymond: Hongchan not here
            - Raymond: Paul, you mentioned compute-pressure in the issue?
            - Paul: Yes.  It's useful to have, but it doesn't replace what
            we want here.
            - Render quantum size
         <https://github.com/WebAudio/web-audio-api-v2/issues/13>
            - Raymond: No updates to explainer but there's a branch for it
            somewhere. Will get it updated before the F2F.
            - Others?
            - Michel: Want to reuse AC graph to OAC to render.
               - Paul: It's issue v2 74
               <https://github.com/WebAudio/web-audio-api-v2/issues/74>.
               - Paul: It's more memory because every node is duplicated
               and all the internal state needs to be transferred.
Sample rate may be
               different too.  Progressive rendering would be really
helpful for DAWs.
               Also want to be able to close OAC.
               - [Raymond failed to capture some discussions]
            - Michel: What about multichannel input from, say, 8-channel
            input?
               - Raymond: This is out-of-scope for WebAudio.
               - Michel: Yes, it's about multichannel input.
               - Raymond: WebAudio should be able to handle it; it's really
               an issue with MediaDevices allowing more than stereo.
               - [Raymond describes some work by Hongchan in Chrome for
               supporting this.  Chrome's implementation is really
hard-wired for stereo
               to support AEC, and noise suppression, etc. for
WebRTC-like use-cases.]
            - Michel: Is there time to present some of the recent work on
            WebAudio?
               - Raymond: There should be some time available either in the
               F2F meeting or later.


See everyone next week for our second virtual F2F meeting!