Re: WebIDL, optional attributes

On Wed, Jun 10, 2009 at 1:55 PM, Marcin
Hanclik<Marcin.Hanclik@access-company.com> wrote:
> Hi Andrei,
>
>>>Yes, I will ask the question about it. Or, are you part of the WebApps
>>>WG? In that case, perhaps you could bring up the issue with them?
> Yes, I already drafted an email about it, but I am waiting for a conclusion of the discussion in this ML.
>
> It would be interesting to know whether the existing implementations are in ES or Java?
> ES Number is 64-bit anyway (so float and double have the same mapping there), but Java makes a distinction.
> The question is:
> Do we want 64-bit or 32-bit values for: latitude, longitude, altitude, accuracy, altitudeAccuracy, heading, speed?
>
> If 32 bits are enough, then I would suggest updating geolocation.
> If 64 bits are required, then WebIDL spec must be modified (I can send the proposal to WebApps after having the conclusion here).
>

With single precision floating point you can represent longitudes with
a precision of about 4 meters. This isn't enough. Please fix WebIDL.

Thanks,
Andrei