> > Yes, this is the kind of approach I was after. Not a new X server, but
> > rather a straightforward extension to existing ones.
> >
> > The video mode programming issue is already black magic, and totally
> > separate from actually drawing stuff on the screen. I can't for the life
> > of me imagine that there would ever be an excuse to make a completely new
> > X server just because the mode programming changed.
>
> In theory its separated, but X servers do are complex things. I don't know
> whether you ever read X server code, but I did.
So have I. And the missing separation between mode setup and drawing setup
in XFree86-3.x is nothing but a design flaw. XFree86-4.x will correct
that.
> > That's like separating the S3 X server into ten different X servers
> > depending on which clock chip is on the board. That would just be silly
> > and stupid, wouldn't it? But it's really almost exactly the same issue:
> > the fact that the kernel handles mode switching could just be considered
> > "a software clock chip" kind of thing. No?
>
> Right! Finally we're getting at the same wavelength :-)
>
> But for the 3.x line, XFree86 has many different accelerated X servers, each
> with their own screen init functions.
Actually, there are only three "serious" separate accelerated X servers
that are worth looking at. Mach64, I128, 3Dlabs. The rest are obsolete
because they are either supporting seriously outdated hardware (8514,
Mach8, etc) or are being superseded by other code (P9000, S3, S3V)
> The current accelerated XF68_FBDev is
> experimental, and we did it the way we did because that was the only way it
> made sense to us (at that time, 4.x is not finished). Don't forget that the
> `normal' XFree86 doesn't have any acceleration support for the NCR77C32BLT nor
> for the IMS Twin Turbo. We borrowed XFree86 accel code for the ATI Mach64 only.
>
> For the 4.x line, things can and should be done The Right Way (I hope).
So do I. I would like to encourage people who would like to see 4.x do the
Right Thing wrt framebuffer devices to join The XFree86 Project and help
to unsure that the separation between mode setup and drawing setup can
support things like fbdev as well.
Dirk
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/