> >But it should be possable to write a driver to take mouse actions
> >and turn them into actions for the rest of the kernel. Like
> >maybe making fake adb events, but I'm not sure if the current adb system
> >would handle mouse events if there's no adb init.

I think we ought to have a two-layer device driver for the mouse, like
the SCSI driver.  /dev/mouse, or perhaps /dev/mouse0, /dev/mouse1, ...
(I thought the same thing about the keyboard, but there are some
design issues to be resolved).  /dev/mouse* can be MI across all the
platforms NetBSD supports.  Then use of non-adb mice on NetBSD/mac68k
might have a meaning.

> But this would allow the use of mouse in the console (to do cut & paste) as
> on Linux x86...

Just FYI, cut & paste is implemented in dt.

> >I thought we had working 3-button mice? Ken Nakata figured out the multi-
> >button protocall, and (I thought) most non-three-button-protocall
> >mice used a fake keyboard for option-left-arrow and option-right-arrow
> >events (mine does).

Besides the mice that support the Apple Extended Mouse Protocol,
MicroSpeed's Mac Mouse Deluxe (?) has been supported, thanks to Taras
Ivanenko.  It seems to me that MicroSpeed is pretty easy to get, at
least in the eastern US.

ken