See also: IRC log
<rtoyg_m> present rtoyg_m
<ghaudiobot> [web-audio-api] rtoy closed pull request #855: Fix #30: allow disabling resampling for decodeAudioData (gh-pages...30-disable-resampling-for-decode-audio-data) https://github.com/WebAudio/web-audio-api/pull/855
padenot, we are starting the call
<ghaudiobot> [web-audio-api] rtoy closed pull request #843: Make HRTF panning k-rate (gh-pages...837-pannernode-a-rate) https://github.com/WebAudio/web-audio-api/pull/843
<scribe> ScribeNick: joe
<ghaudiobot> [web-audio-api] rtoy pushed 1 new commit to gh-pages: https://github.com/WebAudio/web-audio-api/commit/1c3ff09b679ee0ac255476547e88aca7547f09a7
<ghaudiobot> web-audio-api/gh-pages 1c3ff09 rtoy: Merge pull request #843 from WebAudio/837-pannernode-a-rate...
hongchan: remainder of the week
is spent on absorbing Domenic's feedback and some editorial
issues
... also lifetime needs to be discussed. otherwise in a good
place
mdjp: I can see 6 issues from Domenic. Nothing in there is a problem?
hongchan: nothing serious
mdjp: we need to review a PR sometime in the next week
hongchan: I'll have that next week
mdjp: around AudioWorklet -- Domenic's feedback is under control then. what do we need to do to have a review from TAG?
hongchan: still waiting for Alex
but he's too busy. We need to make an official request to TAG
for review
... I'll file an issue on the issue tracker
mdjp: fantastic
joe: I think it should wait for your PR with the arch. changes
mdjp: as soon as you're finished,
please file that issue
... if anything would hold this up, we need to get on that as
soon as possible
... is there anything else around Worklet you'd like to
discuss?
hongchan: lifetime topic, it's
important
... not sure we have to go into huge depth
... we need to talk about the scope of how lifetime is
addressed
<rtoyg_m> https://github.com/WebAudio/web-audio-api/pull/959
https://github.com/WebAudio/web-audio-api/issues/475
<mdjp> joe main difficulty is whether we use the existing mechanism of references or a new method using a call to terminate. The cases of existing nodes which need to deal with retention or GC, a useful approach might be to allow an audioworklet to have control over a reference which retains it
<mdjp> joe the advantage is that if you don't want explicit retention then you will still free the reference when other (input/output) references are removed
<mdjp> joe terminate is also an option. All users are obliged to use it to prevent the node persisting forever. Forces devs to think about and be aware of lifetime.
<mdjp> hongchan if you return nothing? joe then the return would be undefined - any falsy value causes active reference to be removed but other references still exist to keep the node active (connections)
<mdjp> joe if the node goes away process would no longer be called which should be observable.
<mdjp> joe consequence node may not be gc's but the node will not stop proessing as long as n refeerences exist.
<mdjp> rtoyg_m we agreed (tpac) we could have a node that is 10 secs long. If nothing is connected for 10 seconds it will not play as it has already completed
<mdjp> joe if not connected but going to be connected later then something has a reference. The reference keeps it alive.
<mdjp> joe important that if people forget to terminate there is a mechanism to prevent indefinate lifetime.
<mdjp> rtoy return value from worklet could be error prone
<mdjp> joe option flag set in process which can be interrogated on each cycle to determine if code should continue to run
<mdjp> rtoy this less error prone but not necessarily preferred. Maybe disliked less
<mdjp> rtoy specing for things with many references such as worklet it is a good idea to explicitly close them rather than depend on gc
<mdjp> rtoy will follow this up with a google collegue.
<mdjp> joe what happens if we adopt a terminate style api and no one calls it. rtoy - it stays alive forever even if not connected to anything
<mdjp> hongchan you also have to explicitly call terminate to stop a worker. Terminate came from worker joe maybe it is a good thing ian and I trying to move away from worker still thinking about it and this will be a really good topic for TPAC discussion
<mdjp> joe can make an aletrnate PR based around terminate - or can discuss in the issue as a comparison
<mdjp> hongchan TAG may think something different so alternatives would be a good idea.
mdjp: next steps here are TAG
review and working out some kind of approach to lifetime
... great progress going on here
<rtoyg_m> https://github.com/WebAudio/web-audio-api/pull/902
<rtoyg_m> https://github.com/WebAudio/web-audio-api/pull/856
mdjp: let's get input from the list regarding the two issues above
mdjp: next thing is to review the
TPAC agenda
... we want to make this as useful as poss. and we have some
goals for what we want to achieve
... if anyone has any issues/suggestions/proposals for agenda,
please supply
... hongchan, will yo ube able to dial in?
hongchan: I'll definitely try
clilley: did we ask for telecomm equipment?
mdjp: we'll just use Skype/hangouts/whatever
rtoy: one quick comment -- do we need to talk about rechartering?
clilley: there should be an
announcement that the WG has been extended to 1 Dec to allow
the group to complete its work
... also a split between V1 and V2 deliverables
... currently w3c is pushing groups to have an offline mode so
that if someone doesn't get to a call, they can still be
included in decisions
mdjp: I think that's fine on the
assumption that we probably need to get through V1 as quickly
as we can
... I'll slot something in the agenda shortly to discuss
charter etc
clilley: and another thing -- have we had any requests for observers?
mdjp: haven't seen any yet
... we will see everyone on the 22nd
joe: are there other things relevant to acheving CR transition?
clilley: mostly wide review. has there been any interaction with accessibility?
joe: not that I know of
... can you make an introduction?
clilley: yes I'll take care of that
mdjp: in terms of that wide review do we need to prepare any sort of document?
clilley: yes it's called a
disposition of comments. we can show that we have issues that
have received comment and been responded to
... the main thing is not to have dangling non-responded-to
issues or ones where the response made the commenter
unhappy
mdjp: I'm happy to put that together
<mdjp> trackbot, end meeting
This is scribe.perl Revision: 1.144 of Date: 2015/11/17 08:39:34 Check for newer version at http://dev.w3.org/cvsweb/~checkout~/2002/scribe/ Guessing input format: RRSAgent_Text_Format (score 1.00) Found ScribeNick: joe Inferring Scribes: joe Default Present: rtoyg_m, joe, hongchan, mdjp, ChrisL Present: rtoyg_m joe hongchan mdjp ChrisL WARNING: No meeting chair found! You should specify the meeting chair like this: <dbooth> Chair: dbooth Found Date: 08 Sep 2016 Guessing minutes URL: http://www.w3.org/2016/09/08-audio-minutes.html People with action items: WARNING: Input appears to use implicit continuation lines. You may need the "-implicitContinuations" option.[End of scribe.perl diagnostic output]