- From: Harald Alvestrand <harald@alvestrand.no>
- Date: Wed, 19 Mar 2014 10:43:54 +0100
- To: public-media-capture@w3.org
On 03/18/2014 07:40 PM, Jan-Ivar Bruaroey wrote: > On 3/18/14 10:06 AM, Harald Alvestrand wrote: >> On 03/17/2014 05:20 PM, Jan-Ivar Bruaroey wrote: >>> On 3/17/14 2:33 AM, Stefan Håkansson LK wrote: >>>> On 2014-03-15 23:29, Martin Thomson wrote: >>>>> On 14 March 2014 16:24, Cullen Jennings <fluffy@iii.ca> wrote: >>>>>> Two optional constrains, the first one saying the source is A and >>>>>> the second one saying the source is B. >>>>> Or you could try this: >>>>> >>>>> navigator.getUserMedia({ 'sourceId': 'A' }, success, function() { >>>>> navigator.getUserMedia({ 'sourceId': 'B' }, success, failure); >>>>> }); >>>> Hm. Would you not need to push in a "require: 'sourdeId'" in the first >>>> gUM? Otherwise it would be "prefer" and if treated like optional mean >>>> that gUM would succeed even if it could not be satisfied. >>> >>> Correct, things are optional by default, so it would be: >>> >>> navigator.getUserMedia({ sourceId: 'A', require: 'sourceId' }, succ, >>> function(){ >>> navigator.getUserMedia({ sourceId: 'B' }, succ, failure); >>> }); >>> >>> >>>>> It seems like the example is basically contrived, so why not incur >>>>> the >>>>> additional user prompt? >>> >>> Yes, it is hard to judge what's acceptable when we don't root things >>> in real use-cases. >>> >>> SourceId strikes me as the "anti-constraint", i.e. how one subverts >>> the normal process, so we should perhaps not design the normal >>> process around it? >>> >> >> I think it works well as a constraint - sometimes you definitely want >> a specific source, sometimes you would prefer one but can live with >> anything, sometimes you just want the system to pick one. > > Sure, but when would you prefer two? That's the case you brought > forth, and we're asking what the use-case for that is. The only case I can think of that comes reasonably naturally is an XBox, where there might be an external camera (sometimes connected) and a Kinect (which may be represented as two cameras, one with real video and the other one with depth information). I may be eager to avoid the depth "camera"; neither of our proposals have an "any value but..." option.
Received on Wednesday, 19 March 2014 09:44:32 UTC