Keymap issues with Pointer+Keys device

[Peter Hutterer]

On Wed, Sep 24, 2008 at 02:12:48AM +0300, Daniel Stone wrote:
> Maybe I'm missing something, but can we not do as SwitchCoreKeyboard
> used to do, and have all requests for keymap/whatever act on the device
> that last sent a key event to that client?

right, we could. The equivalent now would be SetClientPointer, which the
server would invoke after each key and/or button event.

The reason why I was against it so far is because it only helps in the short
run, and even then only to some extent. It sacrifices openly broken but
predictable behaviour for less broken but unpredictable behaviour.

The correct approach is to have a WM that requests SetClientPointer when
appropriate.

Cheers,
  Peter