Subject: Re: On wscons, mice, and keyboards
To: Lennart Augustsson <augustss@cs.chalmers.se>
From: None <rvb@sicily.odyssey.cs.cmu.edu>
List: current-users
Date: 07/29/1998 15:08:50
Lennart Augustsson <augustss@cs.chalmers.se> writes:

> > > Just to try it out I implemented the "mouse concentrator device".
> > 
> > Why not do this with a userspace process, and a `loopback' character
> > device? That way you can multiplex other things too.
> > 
> > I used to do this with a pipe when I wanted to use both a ps/2 and a
> > serial mouse (before the input extension appeared). The problem with a
> > pipe is that if the daemon dies, the client side gets EPIPE.
> I think that could work pretty well for mice.  A little complication
> in the daemon is that it has to watch for new mice being connected
> (and possibly create new device node if you run out).

I have a feature in my kernels, with old style ps2 mice, that lets
keyboard strokes get mapped into mouse events.  This way I can get all
three buttons on a laptop.  (also the buttons latch which is
occasionally useful.)  It would be nice if this could be a userland
hack that feeds into the mouse concentrator idea.  And ...
you could build a program that tracks mouse events and plays them back