> Maybe I'm missing something basic here, but should there be a 1:1
> correspondence between a wsmouse? and a device?
My point is that today we must have such a correspondence, but
I'd like to get away from it.  The most likely scenario for
a machine with a mouse is that all mice connected to it should
be "in parallel", i.e., should appear as one "device".

If we want to put this functionality in the kernel or in X
is another matter.

>   Why not allow
> a process to open a /dev/wsmouseN and then via ioctl's select
> what mice to get events from.  (the problem with multiple devices
> has to do serialization of events).  The default could be defined
> in the config file be either ALL mice or a device per mouse.
Using an ioctl() to control it is certainly one way.
Maybe just a /dev/wms which just unconditionally joins all /dev/wmsN
not otherwise opened is a simple solution.
If we want it in the kernel at all.

        -- Lennart