> There are two problems with this.  First, your kernel's having trouble
> getting the video card to work.  This makes it impossible for the
> color server to do color because there has to be some device driver
> support for the server to initialize video mode, access CLUT, and so
> on.

I don't think that the color kernels or the SLOTMAN kernels have
support for multiple color cards. Though, if the interrupts aren't being
disabled that isn't good either. Sounds to me as though he's not using
a color kernel at all.

> Second, the color X server doesn't support greyscale (oops!).  It only
> supports B&W and 8-bit pseudo-color modes.  Therefore, even if your
> kernel did support your RasterOps card, X server couldn't run in
> greyscale mode.  I might work on it one of those days, but when, I
> have no idea.

It's funny actually. For most cards, grayscale mode and color mode aren't
really any different. You just set a flag internally and the card mixes
the RGB channels so that they appear as grayscale. The server shouldn't
worry too much about whether the card is in color or grayscale. It should
just worry about whether it's in 8 bit mode or not.

> (Is anybody interested in salvaging my color server source code from
> my disconnected disk drives?  If you are, please contact me privately,
> so we might be able to arrange it to be sent to you.  Oh, you'll have
> to promise me not to peek my private stuff on them ;-)

Just out of curiosity, what kinds of changes did you have to make to the
server? Did you just have to patch in some code to check/change the modes
and change the CLUT or did you actually have to go in and write drawing
routines?

You may want somebody to get your source since we're moving from the old
X server to an XFree86 style server. We've got it running in B&W but
color would be better. I'm also planning on adding hardware page flipping
support calls so somebody might want to add some code to accelerate X
using that.