> this isn't the 8-bit color intvid patch for color X, is it? :-)

Nope. This is the old intvid0 stuff. Intvid0 was my attempt to get more
machine working on black and white video. I thought I submitted those changes
a long time ago but they somehow mysteriously either never made it or
simply disappeared. Though some machines, like the Classic II, did benefit
from the changes when Scott used the code (I think) to make the intvid
tables in machdep.

That was quite a while ago. I was going to make an intvid1 but that was
going to be a unified black and white interface compatible with the
colorkit stuff. I just skipped that and went straight to intvid2.

I'm still working on it now. It's pretty tough to coax color out of
68k Mac hardware. :-) Just when I think I'm almost done I find out something
that changes what I'm doing.

Good example: I thought I was very close to a Quadra video driver minus
resolution changing. I found out a problem I was having was related to
the fact that changing color depths and changing resolution are intrinsically
tied to each other. In other words, I now have to write the support code
for resolution switching which I was planning to put off until after the
first release. Oh well. I guess we'll get it all on the first release :-)

> btw, instead of making it an option in compiletime, if you get the color
> intvid stuff finished, could you look into trying to make it an lkm? i use
> the color video lkm for my nubus card now, and lkm's save time and
> problems of compiling kernels. more organized too, maybe.
> later,

Well, unlike the NuBus video, the internal video of your machine is a true
I/O subsystem. Not supporting it in-kernel would be like not supporting
the serial ports. :)

There are also some changes that have to be made in general that would be
useful even for folks running headless. Automatic detection of the monitor
and turning off interrupts are still very much useful.

Of course, you can argue that just about everything except perhaps network
and disk support should be LKMs. :)

However, the point you are making is valid. I'd like to make it so that
folks can compile a more streamlined version without drivers they don't
need.