multi-touch-dev team mailing list archive

[Mohamed Ikbel Boulabiar <boulabiar@xxxxxxxxx>]

It's a Big Question, and it needs to be resolved in the "right" way.
But the definition of right depends on the researcher/developer view.

putting input in the same wallet as display server is "right" from
developer perspective because this means less processes, so less
context switches etc.



IIRC the last year we discussed input as a way to see a device as a
set of sub input providers you could subscribe to them and modify them
dynamically.
So we have no more a definition of device by it's hardware package.
Routing events to applications becomes easier if a new device should
be supported.

To map this model to real-world it's a more generalized xkb where you
define the mapping between a key and what you want to appear on screen
when typing, and this through a file (which may be compiled when
launching the input server).
Gestures in our case are a block added to the bare touch input.

Changing this needs many modifications to the whole input layer, and
it needs really a lot of time, but once done, whatever new input
device should be easily supported without asking many questions on how
to do that and where to put its special handling code/extension.

Many problem exist now, people still think sensors are not input
devices so have special code, some buttons like the ones emitting
events when there is a screen rotation needs special treatment. etc.
There no sort of good abstraction of from where comes the event and to
where it should be routed. interface details should be hidden in lower
layers.


All of this needs more reflexion, ideas and prototyping.

i

On Tue, Nov 23, 2010 at 4:13 PM, Mathieu Virbel <mathieu@xxxxxxx> wrote:
> 2010/11/23 Duncan McGreggor <duncan.mcgreggor@xxxxxxxxxxxxx>:
>> On 11/23/2010 04:22 AM, Henrik Rydberg wrote:
>>>
>>>> Speaking of wayland - are there any plans on improving the input layer
>>>> of wayland? How mouse
>>>> and touch and gestures input will be handled there?
>>>
>>>
>>> Big question, no answer. :-)
>>
>> There is a rumor around that a certain European researcher is working on
>> a new form of input for Linux... we'll let you know if anything comes of
>> this. (It likely wouldn't be for another cycle.)
>>
>
> It would be very very nice if public can have a little more
> information about that :)
> Like, i'm very scared when i'm reading : "We predict with Wayland the
> window manager will BE the display server and input server". WM act as
> an input server ?