On Tue, 02 Mar 1999 18:39:16 -0800 
 Jonathan Stone <jonathan@DSG.Stanford.EDU> wrote:

 > >- Desktop Bus driver for Personal DECstation
 > >	It can be regarded as flexible two way communication channel
 > >	sending	and receiving packets between host processor, LK501
 > >	keyboard and VSXXX-GB mouse.  This implies 'dtop' driver have
 > >	to be smarter than conventoinal serial connection folklore.
 > >	(But, unfortunately, intermediate driver must emulate LK201
 > >	key sequece anyway)  ADB driver could be modelled.  Exploiting
 > >	IOASIC DMA would be better.
 > 
 > Nope, we dont want to emulate serial devices. There should be an API
 > for passing pointer and keyboard events up to wscons which does *not*
 > require byte-at-a-time parsing, but takes input-pointer updates, and
 > button events, and keycodes in whatever format wscons expects.
 > 
 > If no such API exists, we should create one.
 > (insert annual prayer for Bstreams once again).

Um... PS/2 mouse drivers spew multi-byte protocol packets, right?  Or,
yah, use the macppc ADB code as an example; it does wsmouse and wskbd
attachments.

The point is, you pass entire "events" to wscons... if your device
generates an interrupt-per-byte (like zskbd/zsmouse :-) then that
layer has to buffer them until it has enough info to generate an
event packet for wscons.

        -- Jason R. Thorpe <thorpej@nas.nasa.gov>