Re: Default to an optional/advanced constraint?

Hi Martin,
cc Ningxin,

> On 21 Mar 2016, at 12:18, Martin Thomson <martin.thomson@gmail.com> wrote:
> 
> Two comments:

Martin - thanks for the good suggestions.

> 1. This doesn't seem to warrant any special treatment.  It is a
> definition of the maximum resolution that the device can provide.  If
> someone asks for precision down to 1mm, and the device can't deliver,
> then they don't get that device.

Ok, so let's try to stick with the default behaviour unless someone feels strongly otherwise. This brings up another issue:

Looking at getSupportedConstraints() it appears there's not way to do capability detection based on the constraint value ("mm" or "m" in this case).

The only way to go about that would be to invoke getUserMedia() with this constraint set as mandatory and see whether the call succeeds or fails. However, this requires user's consent.

Have you run into the same issue with any other constraints?

> 2. Why is this precision specified in such an arcane fashion?  When
> talking about measurement accuracy, the normal method is to define the
> size of the error.  Can this "precision" be specified as a continuous
> value with 1== one metre error; 0.001 == one millimetre error.  (If
> you want to advance to entry-level metrology, then we might also talk
> about the probability of error in each measurement.)

Cc'd Ningxin to clarify what level of detail the various underlying implementations expose.

Thanks,

-Anssi

Received on Monday, 21 March 2016 11:59:13 UTC