Re: Proposal: Time out for getUserMedia.

On 10/09/2013 7:10 AM, Harald Alvestrand wrote:
> On 09/07/2013 06:35 AM, Kiran Kumar wrote:
>> Hi,
>> Are there any objections on this proposal.
>> Can we add this to bugzilla/bug tracker, to discuss further and spec 
>> it ?
>>
> With my chair hat on:
> I've created the bug (23194), 
> https://www.w3.org/Bugs/Public/show_bug.cgi?id=23194
>
> But when I review the discussion so far, I have seen a lot of 
> skepticism and "how would this work", and nobody actually speaking out 
> in support of the idea.
>
> It is possible that this is something that should be left for a later 
> version of the spec.
>
> (I'll also note that if I've understood how Futures/Promises are 
> supposed to work, this is much more easily supported, with no special 
> API whatsoever, in a version of getUserMedia that uses futures instead 
> of callbacks. That may argue for leaving it for the proposed future 
> spec version that supports futures.)
>
>             Harald
>
Harald,

     Futures/Promises wouldn't help. Essentially we're asking for the 
ability to cancel an ongoing getUserMedia request. This could occur as 
the result of a timeout, or some external stimuli (server notification, 
SMS, etc).

     I see the following issues:

 1. Should applications have the ability to cancel ongoing getUserMedia
    requests?
 2. If so, should we allow ad-hoc cancellation or limit ourselves to the
    use of timeouts?
 3. Do we have a fingerprinting concern for small timeouts/cancellation
    periods?
 4. If so, what should the minimum timeout/cancellation period be?

     What are your thoughts?

Gili