Re: No-change proposals for WebIDL for constraints

On 16/12/2013 6:11 PM, Harald Alvestrand wrote:
> On 12/16/2013 08:13 PM, cowwoc wrote:
>> On 16/12/2013 1:31 PM, Martin Thomson wrote:
>>> On 16 December 2013 10:05, cowwoc <cowwoc@bbs.darktech.org> wrote:
>>>>    [ {"width": 1920, "height": 720} ],
>>>>    [ {"width": 1600, "height": 900} ],
>>> I hope that those brackets are just a mistake.
>> Yes, they are.
>>
>> I just meant that we should be able to specify multiple constraints 
>> per line, instead of one (as is currently the case).
> Lines have no significance in Javascript (well.... rarely; another 
> strange feature of the language), but the difference between [] and {} 
> is significant, and it is important to use the right one.
>
> Your examples are showing an array each of whose elements contains an 
> array, each of which has an object with 2 key-value pairs. That's a 
> novel structure that I don't see a reason for.
>
> I suggested in the original note that
>
> { optional:
> [
>   {"width": 1920, "height": 720},
>   {"width": 1600, "height": 900},
> ] }
>
> which is an array of 2 elements, each containing 2 key-value pairs, 
> should be a legal constraint value.

Yes, we mean the same thing.

I just noticed that you wrote:

> The browser will try to satisfy as many constraint sets as it can[...]

That's kind of unexpected. Earlier proposals were talking about the 
browser selecting at most one ConstraintSet, not a union of multiple 
ConstraintSets. Doesn't that make it harder for humans to foresee what 
constraints they may end up with? Your proposal is less verbose, but 
also less explicit.

Gili

Received on Tuesday, 17 December 2013 21:26:38 UTC