See also: IRC log
<mdjp> am I hitting the wrong webex, still saying cancelled or ended for me
<padenot> The meeting has been cancelled or ended!
<kawai> same here. The WebEX url looks not working.
<mdjp> yes - same here, joe you tried it yesterday, do you have a different url?
<mdjp> or ChrisL can you give it a poke?
I'm using WebEx successfully
Let me paste the link here
whooops
https://mit.webex.com/mit/j.php?MTID=m7b2688a882ac227c098436ea57a692ac
<mdjp> guess we should update the wiki, I hadn't realised the url had changed.
<ChrisL> I'm on the webex too
I completely forgot about the wiki
<rtoyg_m_> Oh. So no phone?
yes, just use the URL above -- we're all on except you
<ChrisL> +1-617-324-0000 US Toll Number
<ChrisL> Access code: 647 342 611
what Chris said [sorry]
ScribeNick
<mdjp> joe last week passed rtoyg_m_ has started to look into it
<mdjp> no progress to report yet joe
<mdjp> started to look in to it - horror......
<mdjp> joe can be reverse enginered but worth understanding to what level of detail we should define the node.
<mdjp> Is fully defining the behaviour the right thing to do?
padenot: after looking for a while,
<mdjp> rtoyg_m_ for the spec we need to specify enough so that it can be implemented - so everything
padenot: it's not the craziest design, but there are magic numbers everywhere
rtoy: like the polynomials -- I have no idea how Chris R chose those numbers
<mdjp> joe is it worth describing the algo in outline without going into full detail
padenot: maybe that is better,
andthat's the direction I was going.
... describe outlines, like knee region, linear region,
etc.
... so I think that's appropriate. I think specing the actual
polynomial curve is crazy and too rigid
chrisl: how dependent is the sound on these specific coefficients?
padenot: it will sound different, yeah
<jdsmith> What's the link for audio?
rtoy: I have a more radical approach: give up, find a paper that describes a dynamic compressor, work from that, and state that all the existing impls are wrong
jdsmith, see https://mit.webex.com/mit/j.php?MTID=m7b2688a882ac227c098436ea57a692ac
ScribeNick joe
padenot: that's doable, but once you remove all the crazy numers from the code it's not too hard
rtoy: you have to think twice
because no one implements what the spec says anyway, so why not
start with a simpler starting point?
... that way we avoid complexity
padenot: I think I've isolated the 4 parts that need to be speced: pre-delay, gain computation, curve sections...
<padenot> https://github.com/WebAudio/web-audio-api/issues/10#issuecomment-267675496
<mdjp> joe why cant this case be less specific v2 could include a more locked down definition
<mdjp> ChrisL this sounds reasonable - not ideal, but could be a V2 feature
<mdjp> joe there have not been a lot of complaints about the compressor
<mdjp> rtoyg_m_ jdsmith no reports
rtoy: according to Chrome's
stats, .001 of pages use DynamicsCompressor
... that's overall pages
<rtoyg_m_> Oscilllator usage: https://www.chromestatus.com/metrics/feature/timeline/popularity/1573
rtoy: Oscillator numbers are about the same amount
mdjp: I agree that starting over again or a super-specific definition are not realistic and we should do that in v2 if needed
hongchan: clarifying the current implementation will be the v2 tack?
joe: no
hongchan: I don't think we shold work on the current impl any more
[no objections]
<mdjp> joe who is in the best place to take this forwards to resolve this by the F2F
padenot: almost finished AudioBuffer ctor, I could pick it up again
padenot: about to push a branch
padenot: what is the operations
that need to happen on the main thread?
... we need to make sure the script is valid; that's long
... my main concern is that this is a huge, potentially
multi-second operation that blocks the thread
... if we are blocking the main thread this is a problem
... wait for a message to come back
joe: but script validation can be done up front by the importScript()
padenot: so what else do we need
to happen on the main thread?
... I dont think anything else can fail
hongchan: I had some discussions with Domenic and let me go back to the documentation of Worklet
<hoch> https://docs.google.com/document/d/1EazMCrJ0UzbKC2lmqysUq-sQJyvsqLsZ9uqaa7LJX68/edit#
padenot: I think this (the
construction of the AWP while blocking) is feasible
... we want to pre-validate the script so that some of these
steps don't need to be done on the main thread
hongchan: resolution is for 1193 is to use MessagePort to return a message indicating the AWP has/has not instantiated OK
This is scribe.perl Revision: 1.152 of Date: 2017/02/06 11:04:15 Check for newer version at http://dev.w3.org/cvsweb/~checkout~/2002/scribe/ Guessing input format: Irssi_ISO8601_Log_Text_Format (score 1.00) Succeeded: s/now/no/ Present: mdjp hoch rtoyg_m padenot joe jdsmith No ScribeNick specified. Guessing ScribeNick: joe Inferring Scribes: joe WARNING: No meeting chair found! You should specify the meeting chair like this: <dbooth> Chair: dbooth Found Date: 08 Jun 2017 Guessing minutes URL: http://www.w3.org/2017/06/08-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]