Bug#255744: xfree86-common: Mouse not found causes gdm to fail -- this *is* annoying

[Sven Luther <sven.luther@wanadoo.fr>]

On Fri, Oct 21, 2005 at 10:23:32AM +0200, Michel Dänzer wrote:
> On Thu, 2005-10-20 at 12:41 +0200, January Weiner wrote: 
> > 
> > > If the device node doesn't exist, it's probably a matter of making sure 
> > > that the backing kernel module (mousedev IIRC) is loaded.
> > 
> > This is what has been suggested by previous posters to this bug, and
> > which should be done, and which has not been done, and which is the
> > reason why I am writing this rant.
> 
> It's none of the X server's business though.
> 
> 
> > Should I repost the bug to another package?
> 
> IMHO yes. As David pointed out, the only 'bug' in the X server is that
> it doesn't support hotplugging input devices yet, but that's an upstream
> wishlist item that won't be resolved anytime soon. AFAICT the most
> effective approaches to resolve this are:
> 
>       * In the 2.6.12-1-powerpc kernel I'm running, the support
>         for /dev/input/mice seems to be built into the kernel in the
>         first place, avoiding the problem of when and how to load the
>         module. You could ask for the same to be done for the kernel
>         you're running.
>       * You could ask for the udev and/or hotplug packages to load the
>         corresponding module by default on bootup, regardless of whether
>         or not a mouse is connected.
> 
> Either of these has the advantage that even without explicit hotplugging
> support in the X server, you can plug in a mouse after the X server is
> already running, and it Just Works(TM). This isn't true if the X server
> just ignores the failure to open the mouse device.

Does the X configuration per default suggest /dev/input/mice now, or still
proposes /dev/psaux ? 

Friendly,

Sven Luther