[PATCH] Input device properties

Peter Hutterer peter.hutterer at who-t.net
Wed Jul 9 05:55:42 PDT 2008


On Wed, Jul 09, 2008 at 11:51:05AM +0200, Simon Thum wrote:
> thanks a lot! Looks good to me, however one question: As I want to 
> intercept updates to props in dix, am I expected to overwrite 
> SetProperty? Can I rely on drivers not cancelling me (i.e. chaining as a 
> part of the spec) or how would you say this could be done best?
 

not sure if I get what you need to do but - the driver is expected to
set SetProperty, which is called when a client changes a device's
property.

If the property is configured to support a pending state, the client can
only ever update the pending state, it is up to the driver to move it
into the current value. This is taken from randr, I doubt somehow that
many input drivers will need it but let's see.

If the property is not pending (the default), you update the property
directly. However, as you said, the driver may cancel you if the data
is invalid. It is recommended that if a property is unknown to the
driver, the driver should not cancel it and just let it be.

does that help? if not, what is it you need to do?

Cheers,
  Peter



More information about the xorg mailing list