Re: Gyroscope/DeviceMotion events

On 24/08/2010, at 7:38 PM, Steve Block wrote:

>> I suggest extracting gravity out to a separate vector attribute, rather than combining it with
>> acceleration. I expect more people would want to know the direction of gravity, so might as well make
>> it easy for them to get it. Also, it means that can be clearly null while it is being calculated.
> Do you mean add a third acceleration vector for gravity? We can't use
> acceleration & gravity in place of acceleration &
> accelerationIncludingGravity as some devices are unable to determine
> the acceleration without the effects of gravity - that's the
> motivation for the property.

I did mean to use acceleration + gravity in place of acceleration + accelerationIncludingGravity, but I'm fine with the existing proposal too.

So if a device can only return a value with gravity, what must it return for the non-gravity part? null? I assume we want to avoid the client having to process the incoming data with a filter to work out gravity? (users may want to tweak the filter)

>> Along these lines, I wonder if DeviceOrientation should have a flag to indicate whether or not it is real
>> world accurate. For example, some hardware might be able to accurately return orientation changes
>> but not have a compass. This is still extremely useful in games, where they don't necessarily care
>> about real-world orientation, but rather the change in orientation.
> Good point. Would this use case be solved by the rotation rate data
> we've added to DeviceMotionEvent?

Sort of, but the DeviceOrientationEvent is much easier to use. I feel bad because I proposed it, but in most cases the gyro change data is too low-level - all you care about is the current orientation :) I'm sure there are games that will use gyro rotation rates, but it would be very nice to have a flag on the DeviceOrientationEvent.

> If not, such a flag sounds
> reasonable to me.

Great, and thanks for all the updates.

For those following at home, WebKit has started landing some of these features (mostly thanks to Steve himself). I'll begin the process of updating the code to match the recent specification changes.

Dean

Received on Tuesday, 24 August 2010 19:27:37 UTC