On Thu, 03 Feb 2000 16:43:26 +0100, ALTMARK(a)de.ibm.com said:
Valdis asked me in a response to my mail on what RS/6000 model I was
on and what
graphic card was installed. It was a 7043 model 150 with a GXT255P graphic
adapter and he told me that the latter was known to be "brain-dead with colormap
management" and that its successor, the GXT2000P, hopefully fixes the problem.
Well, meanwhile I have such a thing in my RS/6000 - and apparently it doesn't
Damn. Well, I guess I won't ask my boss to buy me one then ;)
Just as a quick test - does the gtx2000 support 16-bit colorplanes?
You can just do a 'xdpyinfo' and see. The ancient GT4x I have in my
home RS6K only has 8 and 24 bit visuals, the gtx255 has 8, 16, and 24.
I'm suspecting the 16-bit visuals may be the issue (but can't prove it
yet).
All this would prove that it's a hardware problem on the
"better" graphic cards,
but ... a) all 21.1 XEmacsen I built on my machine (in the same environment and
with the same options to 'configure') have a nice white-red-yellow icon and b)
that old 21.2-b11 XEmacs which still sits around in my filesystem also does it
right, means: colored.
So this leads me to the conclusion that something must have changed in the way
these icons are displayed since 21.2-b11 (or maybe in a later beta - I haven't
built each and every one in between b11 and b19).
The issue is probably the order colors are allocated in. I'm willing to bet
that somewhere around -b12 throughh -b19, something else got changed (when
did we add the fancy new splash screen logo?) which caused the card to
screw up the colormap when it ran out.
No one else on AIX and comparable hardware seeing this? And if
I'm the only one
- could at least someone of the gurus please tell me what has changed in the
code so I might be able to create a workaround for my "brain-dead" ;-)
hardware?
For what it's worth, Netscape 4.05 through 4.7 and all Tk/Tcl since 7.6 or
so have the same problems getting colors right on a GTX255.
I've been hacking at Exmh the last few days, and I'm sufficiently pissed off
at Tk/Tcl botching it that I'm going to go use a can opener on the Tk source
and see if I can figure out what's going on. Film at 11. ;)
/Valdis