According to mw@eunet.ch:
> > I just downloaded Xamiga24-950305.gz in hope that something had been done
> > about my strange problem with the RetinaZ3/A4000 mousepointer only being
> > displayed at y & 0xff00. But that was only a dream :(. Now, X is very
> 
> If some people with '40 and RZ3 could look into RZ3SetHWCloc(): at the end
> of this function you'll find the following statements:
>         *(acm + (ACM_CURSOR_POSITION+0)) = x & 0xff;
>         *(acm + (ACM_CURSOR_POSITION+1)) = x >> 8;
>         *(acm + (ACM_CURSOR_POSITION+2)) = y & 0xff;
>         *(acm + (ACM_CURSOR_POSITION+3)) = y >> 8;
> now... if ANYONE has an idea why the above statements work for '30ies, but
> will just plain skip the third statement on a '40, *please* tell me...

Yes, I have looked at this, and I am as surprised as you are. I have also
tried to set the low bits of y to a value that is increased for every call
to the function, but no response. That is, the hardware must be doing
something funny. Yet, the software for AmigaDOS works, so ...

> For now, if you want to disable the hardware cursor, change the RZ3 driver
> to return EINVAL on the GRFIOCGSPRITEMAX call. That way, Xamiga24 will know
> that this particular grf-device doesn't support hardware sprites, and will
> use software emulation. TAKE IN MIND THOUGH: panning (i.e. transparnet

Ok, fine! I'm just upgrading to a newer -current anyway, so I'll just put
that in right now.

> scrolling of oversized screens when the mouse hits the edge of the screen)
> works ONLY with a hardware cursor, because it's a feature that is hidden
> from the Xserver and coupled to sprite positioning ioctl's. If you do this,
> I'd suggest you set screen logical size to physical size (look for the
> comment: /* This is the logical ^    ^    screen size */ in grf_rh.c).

Eh, no problem for me, I don't feel any urge to have an oversized screen
when I am already using 1280x1024. Hmm.. Perhaps if I could get a 2x2 display
this way, but then the ram on the board isn't going to be enough
(1.2M*4 > 4Mb), so I think I'll stick with the fvwm virtual display.

> So: in grf_rh.c, in function rh_getspritemax(), instead of returning 0 at the
> end, return EINVAL.

Got it!

> My primary reason to use the sprite is not speed, but transparent panning
> support... (hey, at least *I* am using this feature :-))
> 
> > I have tried compiling my own X-server, but the compilation sucks unless
> > the whole distribution is compiled (I gather). I get lots of errors
> > compiling it, and I haven't even reached to the hardware dependent stuff.
> 
> You *should* be able to compile the server without compiling all of the
> library/other-programs stuff, but there are some dependencies...

Well, I haved changed many Imakefiles to make it work, but there were other
problems with the compilation after the dependencies were through.
But now that the only change needed is in the kernel, I guess I'll skip
the compilation of the server to save me all the trouble.
--

<-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=->
: E-mail: ajo@ludd.luth.se                            Amiga 4000 is it! :
: S-mail: Andreas Johansson, Karhusvagen 5  6:618, 977 54 Lulea, SWEDEN :
:       There are two groups of people I'll _never_ understand:         :
:                     math-professors and women                         :
<-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=->