Blog series on XInput 2 [LWN.net]

Blog series on XInput 2

Posted Jun 19, 2009 8:59 UTC (Fri) by dgm (subscriber, #49227)
Parent article: Blog series on XInput 2

First of all, I just had a quick read of the blog entries, so my opinions are not very strongly based.
That said, I'm not completely comfortable with this extension.
I fail to see it mapping cleanly to multi-touch devices, and I think we are going to see more and more of those. Also, why have pointers and keyboards to be paired?
By the way, "master" and "slave" sounds odd. "router" and "source" would be more appropriate (I think those words are event used somewhere to explain their relationship).

to post comments

Blog series on XInput 2

Posted Jun 19, 2009 10:06 UTC (Fri) by tialaramex (subscriber, #21167) [Link]

Pointers and keyboards need to be paired because "a keyboard and pointer" is the abstraction to which existing applications are written. Of course you could reject this abstraction, but Xi2 isn't about a radical change to the way all the world's GUI applications are written, but just improving the X server.

The _physical_ hardware for this abstraction could be just a touchscreen used with a virtual keyboard. Abstractions are always a bit leaky, but as much as possible the developer shouldn't need to know that I prefer a trackball, and an AT keyboard or if I have an experimental holographic touchscreen interface.

Multi-touch is probably NOT best approached as if the user had two (or three, or five...) mouses connected to USB ports. The same blog has a post about this. When I use two fingers to "squeeze" on a multi-touch pad, that's really not two separate events which just happen to be mediated through a single piece of hardware - it's clearly only one event, but emitting something more complicated than a single (x,y) location. This is nothing new - there have long existed devices which emit (x,y,z) with z pressure, and sometimes also a twist or an additional analogue input (trigger pressure).