dinput: Add mouse warp override config option (resend)

[Vitaliy Margolen]

Roderick Colenbrander wrote:
> Hi,
> 
> Two days ago I spoke with xorg devs (Peter Hutterer / Daniel Stone) in
#xorg-devel. In short Xorg 7.5 (guess that will be out during spring next
year) will feature Xinput2. Xinput2 is on the drawing table now and one of
its features will be relative mouse movement. It is an important feature
which I believe they also need for mouse tablets. They aren't yet sure on
how it will work but one of the ideas is to extend the current events with
relative mouse coordinates OR to add an ability to receive one of both.
> 
> In the current Xorg (7.3) it is possible to open the core pointer /
keyboard as Xinput devices. Though you can't receive events from them. They
said that the originating devices do generate events and that using
DEVICE_CORE you can match them to the core devices or something like that.
> 
> They aren't very happy with the Xorg 7.3 code (the core pointer /
> keyboard
stuff) and recommend to use the current GIT version or to wait for Xorg 7.4.
> 
> I don't know any experience with input on X, so I don't know if the
current code is already enough for Wine but I fear we need to wait for Xinput2.
> 
> Regards,
> Roderick 
> 
>> Stefan Dösinger wrote:
>>>> That's assuming XI2 will be able to send actual mouse (other devices)
>>>> movements to requesting applications. Right now nothing can (DGA
>>>> doesn't
>>>> count it's not available anymore).
>>> Can you get in contact with the Xorg mailing list on that issue? As far
>> as I
>>> understand the XI2 plans raw input processing is planned and a high
>>> priority(not only for wine, but also to layer DGA emulation on top of
>> it). I
>>> barely know the details, and since you worked on this it would be very
>>> helpful if you talked to the Xorg guys
>>>
>>>
>> Which mailing list? Why don't you cc this message there as well?
>>
>> Vitaliy.
>>
> 

I've played with older version. And I don't think there are anything new in
the new version other then ability to open core devices. Which is impossible
in the older versions.
However they still generate events with absolute coordinates. While device
query says it's in relative mode...

Honestly I don't care absolute or relative - as long as events:
1. Produce events from the _device_ (not pointer)
2. Don't get stuck in the corder (see #1)

Vitaliy