F2F 2021 Minutes for Mon May 10

Here are the minutes from today's F2F meeting:

Attendees


*Paul Adenot, Jack Schaedler, Philippe Milot, Raymond Toy, Hongchan Choi,
Christoph Guttandin, Matthew Paradis, Chris Lilley, Chris Wilson*Minutes

   -

   16:30-17:00 UTC (9:30-10:00 am PDT): Current status of V1, living
   standard
   -

      ChrisL: Proposed recommendation in place, 5 approvals already
      -

      Matt: Any opposing?
      -

      ChrisL: Not expecting anything.  Still waiting for Mozilla’s
      testimonial.
      -

      Paul: Working on it.
      -

      ChrisL: How about you Ray?
      -

      Ray: Working on it.
      -

      Chris: What uses webaudio? iPlayer?
      -

      Matt: Not yet.
      -

      Hongchan: Google Meet, Stadia, SoundTrap, Amped Studio, BandLab,
      BeatPort, Soundation, Zoom
      -

      ChrisL: Spotify
      -

      Raymond: What the state in browsers
      -

      Paul: Intend to implement everything in the spec.  suspend/resume is
      biggest missing feature with WPT.
      -

      Paul: Missing some AudioParam items too.
      -

      Hongchan:  Mostly complete for v1.
      -

      Raymond: Missing output Latency and MediaStreamTrack
      -

      Hongchan: Working on fixing worklet test failures
      -

      Philippe: Where can I look up impl status?
      -

      Raymond: https://wpt.fyi/webaudio for one
      -

      Paul: Also
      https://webaudio.github.io/web-audio-api/implementation-report.html
      -

      Jack: I find it somewhat weird that they block the whole site due to
      the lack of WebMIDI. That app can still do a lot without real-time MIDI
      input.
      -

      Raymond: What does that mean?
      -

      Hongchan; Soundtrap says to download chrome to get midi
      -

      Paul: Starting to find resources for MIDI.  Not sysex (but written
      already).  Optimistically maybe this year.
      -

      [Paul describes issues with testing support.  No “fake” device so
      testing permissions is hard.]
      -

      ChrisL: wpt lacks "fake device" support so hard to test MIDI
      -

      Christoph: https://www.npmjs.com/package/midi-test may be helpful;
      I’ve tested it.
      -

      Paul: It would be nice in the browser tests to say add 2 webcams and
      2 midi devices to test these items.
      -

      Raymond: No midi wpt tests?
      -

      Paul: Yes.  Basically just IDL tests since that’s all that can be
      done.
      -

      Raymond: We have 30 minutes before WebMIDI
      -

      Paul: Some topic we can squeeze in?
      -

      Raymond: Let’s try Change outputChannel dynamically:
      https://github.com/WebAudio/web-audio-api-v2/issues/37
      -

      Paul: My last comment still holds.
      -

      Raymond: Without this, can’t polyfill nodes with worklet, like
      delaynode
      -

      Paul; I wonder how VSTs do this.
      -

      Hongchan: Variable output channels? I don’t think VST allows that.
      -

      Paul: VST3:
      https://steinbergmedia.github.io/vst3_doc/vstinterfaces/classSteinberg_1_1Vst_1_1IAudioProcessor.html
      -

      Hongchan: Don’t understand how that can work after construction.
      -

      Hongchan: But the user needs to click on something to make the
      change. Not quite what is requested here.
      -

      Raymond: Could add API to allow changing config, just like how the
      constructor has options to set up output config.
      -

      Hongchan: Possible.
      -

      Honghcan: Maybe leave it but reduce priority from 1 to 3.
      -

      Jack: It's not blocking us, for the reason that you mention,
      Hongchan... if it can happen as the result of a relatively major user
      interaction, it's not such a big deal for many cases.
      -

      ChrisL: Does that mean you can’t polyfill a gain node?
      -

      Raymond: That should work, but you can’t do a delaynode correctly.
      -

      ChrisL: Ah, because the change should happen after the delay.
      -

      Raymond: Chrome does this wrong, and no one has complained about it.
      :-)
      -

      Raymond: I think browsers have everything available to do this
      -

      Paul: Yes, that’s true.
      -

      Raymond: but it’s really complicated and things will apply next time
      at the earliest.
      -

      Raymond: I don’t think it’s really worth adding this complexity to
      the worklet.
      -

      Jack: Yeah, I definitely see how the conceptual issue makes it a P1.
      -

      [Conclusion: Reduce priority but leave it open]
      -

      ChrisL: Can we get a minuted resolution to make the cookbook a W3C
      Note?
      -

      [Raymond: explains issue]
      -

      Resolved: publish cookbook as Note
      -

      [ChrisL: Explains what happened when the IETF server went down.]
      -

   17:00-18:00 UTC (10:00-11:00 am PDT): WebMIDI
   -

      ChrisW: Not a lot has happened since last time; nothing really
      critical. Usage continues to grow.  Want to update security to require
      permission.  Need to push out a spec
      -

      Paul: (As mentioned)  Getting  resources, and everything still works
      with old tests.  Sysex is a separate question that is a matter of policy
      not impl.  Would be nice to have a second impl.
      -

      ChrisL: https://bugzilla.mozilla.org/show_bug.cgi?id=836897#c78
      -

      ChrisW: I think the only issues should be in v2.  What isn’t are
      minor testing issues needed for V1.
      -

      ChrisL: Can we add some device (as mentioned above) for testing?
      -

      ChrisW: Could implement virtual loopback devices, but that’s quite
      challenging.  Don’t want to allow loopback between threads. (Cross-thread
      messaging is bad.)
      -

      ChrisW: Does Mozilla want to do security and privacy review again?
      -

      Paul: Did that a while ago, and nothing’s really changed. Sysex.
      -

      Raymond: Safari?
      -

      ChrisW: Nothing new.  Implementation pressure might help if Mozilla
      implements it.
      -

      Paul; I think we filed all the issues we found so DOM work and
      interfaces are done.
      -

      Jack: It's a bit of a chicken and egg situation. Now that Worklet is
      there on Safari/iOS, I think more people are going to be willing
to invest
      more on WebAudio and those WebMIDI numbers will continue to grow.
      -

      Hongchan: Yeah. That's a good point.
      -

      Jack: Have the Safari folks ever said if there's something particular
      about WebMIDI that makes it fall into the "harmful" category? Like, would
      they be willing to implement if it were behind a permission, with no
      sysex?  Or is the "harmful" marker just a proxy for something
more similar
      to what Chris was discussing off the minutes?
      -

      ChrisW: Safari is generally opposed to HW access.
      -

      Jack: Yeah, I can understand the reluctance.
      -

      ChrisL: OSC has not really taken over from MIDI like some predicted
      -

      Jack: These folks are doing awesome stuff with WebMIDI: It's also
      very high profile at the moment: https://isartum.net/
      -

      Raymond: What is OSC?
      -

      ChrisL/Hongchan: Open Sound Control
      -

      ChrisL: MIDI 2.0 has a capability negotiation, but I haven't seen any
      actual hardware with it.
      -

   Other business
   -

      Paul: I may have a conflict on Wed.  Will let everyone know by Tue.
      May want to reorder the agenda if so.
      -

      Raymond: Should be fine.

Received on Monday, 10 May 2021 18:11:03 UTC