Re: Ideal? (was Re: WebIDL-compatible syntax compromise)

On 04/02/2014 06:05 PM, Jan-Ivar Bruaroey wrote:
> On 4/2/14 11:28 AM, Harald Alvestrand wrote:
>> the typical "I must have a size in this range but would really prefer
>> that
>> size" example could be expressed as
>>
>> constraints = {
>>     required: "width",
>>     width: {min: 230, max: 1024},
>>     advanced: [{width: 640}]
>> }
>
> This is no worse than in the existing spec, but yes, you probably hit
> the simplest need for advanced right there, since we can't write:
>  
> var constraints = {
>    required: "width",
>    width: {min: 230, max: 1024},
>    width: 640, // error: object property repeat!
> };
>
> Would people be open to (re)consider the 'ideal' extension?
>
> var constraints = {
>    required: "width",
>    width: {min: 230, max: 1024, ideal: 640},
> };

I'd be open to reconsidering that once we've declared consensus for a
single proposal and put it into the document..... I see some nice things
about "ideal", but regard it as a separable concern.

>
> It might let us get rid of getNativeSettings() as well, as you could
> get native info in it from getCapabilities().

I don't think we can - if we have a camera that can be opened in either
640x480@60 mode or 1024x768@30 mode (physically different control
signals sent to the camera for the two cases), getCapabilities() won't
tell us which of these modes it's currently opened in.

The argument I remember sounding realistic for getNativeSettings() (I'd
prefer it to be track.source().getSettings(), but that is a different
discussion) is that if we're sourcing a 640x480 stream off a camera,
we'd want to find out if we could increase to 1024x768 without
reinitializing the camera (or similar changes that could either be
disruptive or non-disruptive, depending on the camera's settings) or
not. I'm not saying I find the argument terribly compelling, but I think
that's the argument.

>
> .: Jan-Ivar :.
>


-- 
Surveillance is pervasive. Go Dark.

Received on Wednesday, 2 April 2014 16:22:51 UTC