Colour problems with latest DIB fix

[Johan Gill]

Since the big DIB fix, red and blue are exchanged in some cases,
for example in the install program for The Sims.

This happens in both depth 16 and depth 24.

/Johan Gill, [e-mail]

[François Gouget]

Johan Gill wrote:
> Since the big DIB fix, red and blue are exchanged in some cases,
> for example in the install program for The Sims.
>
> This happens in both depth 16 and depth 24.

Could you give me some more information about your X server (version,
driver)?

Also, could you compile the following test program, and compare the
log to the attached nt4.log (or send the log to me).
   To compile the application just do the following:
 * I assume that PATH and LD_LIBRARY_PATH are set so that wine and
winemaker will work
 * save dib.c in a directory on its own
 * winemaker .
 * ./configure --with-wine=<where your wine is>
 * make
 * ./dib >dib.log

Note that if you compare the above log generated on a 16bpp server
with nt4.log, you will notice differences. This is because when X is in
16bpp there is some information loss. So it's best to do the test in
24bpp mode.

Another log that would be very useful is:
   ./dib -debugmsg +bitmap >dib-trace.log 2>&1

In fact both are pretty much needed to debug this kind of thing.

[François Gouget]

Johan Gill wrote:
> Since the big DIB fix, red and blue are exchanged in some cases,
> for example in the install program for The Sims.

Actually this may also be a problem with 1, 4 or 8 bit DIBs which the
test application I sent to you does not test. In that case a -debugmsg
+bitmap would tell us if one of these cases is being invoked.