xf86-input-synaptics:master: 1 commit(s)

[Peter Hutterer]

On Wed, Sep 10, 2008 at 04:57:57PM +0200, Sascha Hlusiak wrote:
> Hi Peter,
> 
> > Commit against master at e622b00f...:
> > commit c57a7b463fb86d065fc6fe316ed25f302d51e5c6
> > Author: Peter Hutterer <peter.hutterer at redhat.com>
> > Date:   Wed Sep 10 23:37:27 2008 +0930
> > 
> >     Claim that we are a XI_TOUCHPAD, not a mouse.
> >     
> >     If this still breaks with KDE, fix KDE or the server.
> Say, what is this type_name for? In what client program is it used, how
> did it break KDE?

>From the XI protocol specs:
---
The type field is of type Atom and indicates the nature of the device.
Clients may determine device types by invoking the XInternAtom request passing
one of the names defined in the header file XI.h. The following names have
been defined to date:

MOUSE TABLET KEYBOARD TOUCHSCREEN TOUCHPAD BUTTONBOX BARCODE KNOB_BOX TRACKBALL 
QUADRATURE SPACEBALL DATAGLOVE EYETRACKER CURSORKEYS FOOTMOUSE ID_MODULE
ONE_KNOB NINE_KNOB
---

So type is simply a "standardised" identifier. I'm not sure how popular e.g.
"barcode" or "nine_knob" are, but differing between mouse, touchpad,
touchscreen, etc. is quite useful.

This change was triggered by Red Hat Bug 324721 [1], where we need
gnome-mouse-properties to differ between mice and touchpads. This is the
mechanism in place for it.

Regarding the KDE reference, the diff says:
-    local->type_name               = XI_MOUSE; /* XI_TOUCHPAD and KDE killed the X Server at startup ? */
+    local->type_name               = XI_TOUCHPAD;

How exactly this was a problem I don't know, but we should fix the bug in KDE
or the server (whichever appropriate) rather than work around it in the
driver.

Cheers,
  Peter

[1] https://bugzilla.redhat.com/show_bug.cgi?id=324721