> The short answer is: they could be made to work, but it requires that we
> kludge the alpha to look like a PeeCee. 

I tried to run Linux/XFree86 on Alpha around Linux's 2.1.20 days, with a
ZLXp-E3, a Matrox MGA2164W, an ATI Mach64 and an S3 Trio64.  I was very
persistent, but it never worked consistently.  There were locks on boot
seemingly related to that ``BIOS Emulation'' nonsense, dying X servers,
and outright crashes.  Maybe it works a little better now, but I honestly
don't care.  if some NetBSD developers believe Linux did not make the
so-called Right Choice here, I'm ready to believe.

Maybe a more productive way to expend short-term work (if there is to be
any) would be in getting wscons/fb support for a couple common IC's on VGA
cards, similar to the existing TGA support?  Support for one or two chips
is an acceptable stopgap solution until XFree86 starts taking non-i386's
seriously:  maybe the card in your hand would not work, but a cheap
commodity card within arms' reach would.

It's not a kludge, or a condescention to i386/XFree86.  And if XFree86
improves, the code isn't discarded.

Since macppc's use Mach64's and S3's (?), also Amiga, Atari?, and many
arch's have PCI now, the code could be reused, yes?  And, direct chip
programming would make wscons useful with now-common surplus
fixed-frequency monitors.

Most critically, it brings the Project closer to Andrew Gillham's
1999-04-11 resurrection of the dream which has, I think, been haunting our
collective subconscious since the death of VME ``console cards'', the
Hexostation:  a $2000 box with built-in Eternet, USB, and six cheapass PCI
video cards.

Turning old wintel junk into an X terminal is best described as cute:
useful, cool (in relative terms), but still extremely nauseating.

The Hex-O-Station is something else entirely.  I think even Ronna would
have to say, ``Nice.  ...  so, do you really want to buy that
fifteen-pound bag of Tang, or not?''

-- 
Miles Nordin / 1-888-857-2723
555 Bryant Street #182 / Palo Alto, CA 94301-1700