Re: CfC: approve Errata for Geo v1

On mer., 2014-02-12 at 11:38 +0000, Michael van Ouwerkerk wrote:

> I think the timeout member of PositionOptions should not have a
> default value. That way, it can be 'not present' to replace the
> previous use of Infinity as default.

Good catch, removed.

> 
> On re-reading, I noticed that for maximumAge the spec already
> specifies a default value of 0, so we actually cannot use the 'not
> present means Infinity' trick.
> 
Hmm… True.

> I'd suggest we cannot rephrase step 2 as I suggested earlier, so I
> withdraw that suggestion.

Removed.

>  However, in section 5.2 for the PositionOptions interface there is
> this sentence:
> "Setting the maximumAge to Infinity must determine the implementation
> to return a cached position regardless of its age."

> Since it is not possible according to WebIDL to set an integer type to
> Infinity, how about just deleting this sentence?

I have added that removal in the proposed errata.

>  Not sure how all browsers have implemented this. In Chrome, I can
> specify Infinity in js, and I get a cached position.

That means that at least Chrome is not bound by the "Infinity"
conversion to 0 that my reading of WebIDL imposes; we probably need to
check what other browsers do to understand what the right approach is,
and document it properly in the errata. I'll take a stab at it.

>  In practice, setting a large number is very likely to have the
> desired effect. Using the unsigned long type, a maximumAge of nearly
> 50 days can be specified. If [Clamp] is used, there should be no
> overflow surprises.

Right; but it remains that the spec says something, and at least one
implementation doesn't do it, so we need to figure how to align them and
document that alignment.

Dom