<trackbot> Date: 17 April 2014

<cwilso> I'd like to add FTF meetings to agenda.

<kawai> IPcaller may be me.. but I can not hear any sound....

<scribe> ACTION: Olivier to migrate demo-list to HTML, split learn/showcase [recorded in http://www.w3.org/2014/04/17-audio-minutes.html#action01]

<trackbot> Created ACTION-99 - Migrate demo-list to html, split learn/showcase [on Olivier Thereaux - due 2014-04-24].

<scribe> Agenda: http://lists.w3.org/Archives/Public/public-audio/2014AprJun/0021.html

<Domenic_> constructability should be before subclassing

olivier: start with layering

Domenic_: want to convey idea
... if we were to implement audio tag, how would it be built
... raises questions about what the audiocontext is
... also practical, help web authors have clarity
... there are a lot of edge cases in the interaction between 2 apis

olivier: understand that by doing it we would unearth issues

Domenic: https://github.com/WebAudio/web-audio-api/issues/257

cwilso: Alex and I (and Ray) have had discussion about this
... good exercise to go through

lots of detailed questions that need to be answered

cwilso: question of whether we could use Jer's demo
... serves to demonstrate the complexity of it
... not sure we will build the library

Domenic_: not necessarily your responsibility
... questions raised - playback rate, codecs

cwilso: questions we absolutely have to answer is how the 2 work together
... with the default media element context

Domenic_: some of it may be answered intuitively, but need to be clarified

cwilso: complex to explain everything in terms of javascript with raw access to input and output
... would need to look deeper at implementation

Domenic_: we need better stories for the platform as a whole - related to such discussion and pitfalls
... gap analysis is probably the ultimate goal for this layering discussion

olivier: worth diving deeper now given key people missing (Alex, Paul)

[agreement not to]

** Promises - https://github.com/WebAudio/web-audio-api/issues/252

cwilso: about the ability to use promises
... talking about promise-compatibility
... doesn't seem extremely hard, no large risk
... if we make the change implied by snippet, you won't know whether implementation has changed to support promise

Domenic_: given release cycles of major implementations

cwilso: note Safari has longer release cycle
... but think it's right thing to do

olivier: no way to test/know? (short of versioning)

Domenic_: you could detect this pretty easily
... and shim the promise version

olivier: sounds better. More of a case of "just do it"
... want to know whether this is really v1... worried of scope-creeping ourselves to death

cwilso: we need to fix decodeaudiodata anyway
... might as well make this change

Domenic_: would be happy to sunbmit patch to the spec
... to get started

cwilso: would like to talk with you more about how to incorporate progress update etc

Domenic_: there was also offlineContext.startRendering()

cwilso: would suggest splitting those issues

<Domenic_> https://github.com/w3ctag/promises-guide#one-time-events

Domenic_: happy to create 2 PRs and have discussion in there

cwilso: does that render the event superfluous

Domenic_: wouldn't that be backward-incompatible

cwilso: bigger question

olivier: doesn't sound like it would be necessary - we can reassess if anyone objects to keeping it

** Constructibility

Domenic_: from a JS point of view it is strange that objects come to view without having been constructed
... we have a concrete proposal in https://github.com/WebAudio/web-audio-api/issues/250
... also allows subclassing
... most browsers don't have support for it in webidl yet, however
... should be kept in mind
... constructors should be first

olivier: purely adding constructors - so backward compatible

Domenic_: yes, and behaviour would be exactly the same

cwilso: subclassing - agree, next big topic that needs to be solved
... given that the idea is to allow creation of complex nodes
... still on the bench about constructors though
... a number of the nodes can not be constructed usefully without a context

Domenic_: context would be required

cwilso: don't care that deeply, would probably just use create methods

Domenic_: we know there is C++ magic
... but javascript programmers would expect this to be logical

olivier: slight worry at confusion caused by having the two ways of doing something

Domenic_: developers would be happier to pass attributes to constructor

olivier: this sounds like a CfC

<scribe> ACTION: olivier to organise CfC for constructors [recorded in http://www.w3.org/2014/04/17-audio-minutes.html#action02]

<trackbot> Created ACTION-100 - Organise cfc for constructors [on Olivier Thereaux - due 2014-04-24].

Domenic_: there is also an event that is not constructible

<Domenic_> OfflineAudioCompletionEvent

Domenic_: note that it is absolutely necessary for subclassing

** Subclassing

Domenic_: in general designing for subclassing means being very explicit
... about implementation-specific details
... right now if you try and apply a node method on arbitrary object it will fail
... no subclassable object in web platform right now
... but clearly a big use case for web audio

<Domenic_> clarification: "implementation-specific" isn't quite right; "internal" is more what i meant

cwilso: related to our issues with scriptprocessornode
... we would have to make it a first class citizen first
... everything would have to be describable
... in javascript with input/output stream
... e.g convolution
... can be very difficult to describe thus

olivier: unsure whether we need to release a well-tested v1 before tackling this, sounds like a next big thing

<cwilso> cwilso: I think we need to tackle this up front.

f2f meetings

olivier: have responded to survey, we will be at TPAC
... have asked to be early in week

<scribe> ACTION: olivier to check our preferences for TPAC are for nor overlap with AB [recorded in http://www.w3.org/2014/04/17-audio-minutes.html#action03]

<trackbot> Created ACTION-101 - Check our preferences for tpac are for nor overlap with ab [on Olivier Thereaux - due 2014-04-24].

<Domenic_> what's the date?

olivier: cwilso, go ahead and re-offer to host one meeting before that

cwilso: could simply be an editor get-together

<scribe> ScribeNick: olivier

<scribe> ScribeNick: olivier

This is scribe.perl Revision: 1.138  of Date: 2013-04-25 13:59:11  
Check for newer version at http://dev.w3.org/cvsweb/~checkout~/2002/scribe/

Guessing input format: RRSAgent_Text_Format (score 1.00)

Found ScribeNick: olivier
Found ScribeNick: olivier
Inferring Scribes: olivier

WARNING: No "Present: ... " found!
Possibly Present: Domenic Domenic_ Doug_Schepers IPcaller ScribeNick aaaa aabb aacc audio clarification cwilso has heath https inserted johnwbyrd joined kawai kawai_ mdjp olivier paul___irish rtoyg rtoyg_ shepazu slightlyoff toyoshiAw trackbot
You can indicate people for the Present list like this:
        <dbooth> Present: dbooth jonathan mary
        <dbooth> Present+ amy

Regrets: Giri
Agenda: http://lists.w3.org/Archives/Public/public-audio/2014AprJun/0021.html
Found Date: 17 Apr 2014
Guessing minutes URL: http://www.w3.org/2014/04/17-audio-minutes.html
People with action items: olivier