[RFC] MPX XInput Protocol extensions - detailed nits

[Peter Hutterer]

Daniel Stone wrote:
> On Tue, Mar 04, 2008 at 06:28:17PM +1030, Peter Hutterer wrote:
>> With a bit of effort, we might be able to sort out touchscreen support 
>> by then and add it as a standard to XI2.
> 
> What are we mainly missing, just calibration?  I think this is one of
> the things where the solution is trivial once we actually know the
> problem space.

experience, basically.
The way how the blob events are designed ATM is so general that no real 
device fits the bill.

If the perfect hardware existed, everything would be nice and dandy. it 
doesn't, so we have to figure out what to do.

If you put two fingers down: does it count as two separate blobs? If 
it's the two index fingers for example, yes, I guess it should.

But if you put index and middle finger down close to each other? Two 
blobs? or one big bounding box with the two blobs inside?

This is important, because the hotspot of a blob decides which client to 
deliver the event to. And AFAIK we can't get this data from any device. 
Any IR screen can give you blobs but not identify them. Any 
capacity-based screen gives you identity, but only of the user, not of 
the touch-point.
So, the best you can get at the moment is separate blobs for each finger.