As recently as yesterday, I've had apprunner crash my Xserver and do the colormap-flash thing. I had never seen this until yesterday. Akkana reports seeing this on Linux too.

Didn't pavlov check in some code in the last few days? -pn

Chris: Do you want this back? I see some changes in widget land that have to do with invalidating rects, which might force new paints. There was also some checkins to gfx/src/gtk for printing, warning fixes and "freeing Pixmaps".

This is now only the colormap flashing bug. Xserver crashing bug has been filed: 4076

Could someone charactarize how bad this is? We need to set priority and target milestone.

colormap flashing. When the mouse is outside of the apprunner window, the Entire Screen swaps in a different color map and changes the color for all non-apprunner windows to some random color scheme. Move the mouse into the apprunner window and the exact opposite occurs. This is really annoying, and is the equivalent of running Netscape 4.5 with the -install flag.

I agree this is annoying, but i say who cares. Maybe somebody at sun could help us by looking at gdk ? All we are doing is calling gdk and gtk api functions to render stuff. All color allocation is controlled by gdkrgb. In the pc (linux) world, you can buy a true color vga card for 50 bucks. So, I say this is not a big deal. Of course if you have one of those fancy sun workstations that only have 8 bit displays, you are screwed. So, maybe those nice folks at sun can take a stab at gdkrgb.

I agree with Ramiro.

Pushing this off a few milestones.

This is also happening on sparc-linux. re-summarizing.

HELP WANTED: I need some 8-bit xlib/gtk help for this one. Solaris hardware is complaining, not urgent.

You might take a look at <http://bul.eecs.umich.edu/~crowej/sunfaq/ColormapFAQ.html>. This describes the problem and links to a module that is wedged in front of one of the X color-allocation routines. For best results the module has to be manually incorporated into the user's X environment and used by all X apps, so the simplest strategy may be to build & ship the module, but treat it the same way the old movemail utility was treated.

I already checked in a fix for this. The bug was that Gdk's RGB module will force us to use a private colormap on 8-bit displays. See my fix for bug number 5890. *** This bug has been marked as a duplicate of 5890 ***

Rubber-stamping as verified based on casual reading of both bugs. Mr. McAfee, please re-open if I've missed something. Thanks!