Re: TAG feedback on Web Audio

My deep apologies for not replying sooner. Moving continents got in the
way. Hopefully it's not too late to be constructive.


On Fri, Jul 26, 2013 at 5:41 AM, Ehsan Akhgari <ehsan.akhgari@gmail.com>wrote:

> On Fri, Jul 26, 2013 at 3:46 AM, Olivier Thereaux <
> Olivier.Thereaux@bbc.co.uk> wrote:
>
>> Dear all,
>>
>> Let me start by thanking the TAG for this review. It is great to see the
>> TAG getting its hands dirty and reviewing several specs in the web
>> platform for consistency and quality. I am very grateful for the work.
>>
>> ROC pointed to the following part of the review, which I also find
>> problematic.
>>
>> On 26/07/2013 02:16, "Robert O'Callahan" <robert@ocallahan.org> wrote:
>> >In the discussion of data races, it's not completely clear to me whether
>> >the current situation (as implemented in Webkit/Blink) is considered by
>> >the TAG to be "impermissible visible data races". Can you clarify that?
>>
>> While the issue is obviously controversial in the WG, and the input of the
>> TAG more than welcome, I feel I have to point back at my earlier message
>> on the topic:
>> http://lists.w3.org/Archives/Public/public-audio/2013JulSep/0245.html
>>
>>
>> Language such as “it is impermissible for Web Audio to unilaterally add
>> visible data races” without any consideration of the probability and
>> impact of said data races is, IMHO, rather peremptory and thus ultimately
>> unhelpful in helping our group decide the best course of action.
>>
>
> I already posted my take on this here: <
> http://lists.w3.org/Archives/Public/public-audio/2013JulSep/0257.html>;
> I'd be curious to know what Alex and others of TAG members think about this.
>

My view is that your basis for objecting to data races is *also* legitimate
but ultimately less compelling (to me) than the invariant breakage about
current JS execution behavior.

The argument boils down to this: if the goal of this system is to describe
a low-level system which, at bottom, is about JS driving audio hardware, we
need to design it in such a way that it's sane with respect to JS and
there's no *independent* way for a JS context to have introduced this
behavior visibly, so from the perspective of de-sugaring, this is a visible
side-effect of non-self-hosting, and therefore deeply problematic. It
demands that there be "magic" in the platform. This is an anti-pattern.

But please note that there is no good reason to assume that the position of
> data races not being OK is the weaker position which needs to carry the
> burden of proof.
>

A good point = )


>  One could easily turn the table around and ask for the proponents of
> exposing data races to the Web to bring reasons to justify their position.
> Given the fact that the argument to the contrary isn't very convincing to
> our side, I think trying to put the entire weight of proof on the shoulders
> of only one of the sides of this argument can be unhelpful.
>
> Cheers,
> --
> Ehsan
> <http://ehsanakhgari.org/>
>