Re: Weird color problem with GNUstep/Etoile

[Fred Kiefer]

Andreas Höschler wrote:
> Hi Fred,
> 
>>>> Could you please find out, which colour mode you are using and if you
>>>> hit this specific case? Enabling debug output for "back-art" should at
>>>> least provide some of the data. (Start your application with
>>>> --GNU-Debug=backart)
>>>
>>> I tried
>>>
>>>     /opt/GNUstep/Local/Applications/Typewriter.app/Typewrite
>>> --GNU-Debug=backart
>>>
>>> The application got started with the color problem, but nothing of value
>>> was logged!? A typo?
>>>
>>
>> Yes :-)
>> You should have used
>> /opt/GNUstep/Local/Applications/Typewriter.app/Typewrite
>> --GNU-Debug=back-art
>>
>> I had it correct in one place and wrong in the other.
> 
> OK, here it comes!
> 
> 2007-08-03 19:15:32.946 Typewriter[2803] Initializing libart/freetype
> backend
> 2007-08-03 19:15:32.952 Typewriter[2803] Loading 3 user defined AppKit
> bundles
> 2007-08-03 19:15:32.953 Typewriter[2803] Loaded
> '/opt/GNUstep/System/Library/Bundles/Camaelon.themeEngine'
> 2007-08-03 19:15:32.961 Typewriter[2803] Loaded
> '/opt/GNUstep/System/Library/Bundles/EtoileMenus.bundle'
> 2007-08-03 19:15:32.967 Typewriter[2803] Loaded
> '/opt/GNUstep/System/Library/Bundles/EtoileBehavior.bundle'
> 2007-08-03 19:15:32.978 Typewriter[2803] XIM: '(null)' is not supported
> 2007-08-03 19:15:32.979 Typewriter[2803] Unable to initialize XIM, using
> standard keyboard events
> 2007-08-03 19:15:32.979 Typewriter[2803] artcontext_setup_draw_info
> masks=(ff000000 00ff0000 0000ff00) bpp=32
> 2007-08-03 19:15:32.979 Typewriter[2803] got t=8
> 2007-08-03 19:15:32.979 Typewriter[2803] gamma=1.4
> 2007-08-03 19:15:32.983 Typewriter[2803] artcontext_setup_draw_info
> masks=(000000ff 0000ff00 00ff0000) bpp=32
> 2007-08-03 19:15:32.983 Typewriter[2803] got t=5
> 2007-08-03 19:15:32.983 Typewriter[2803] gamma=1.4
> 2007-08-03 19:15:32.993 Typewriter[2803] artcontext_setup_draw_info
> masks=(ff000000 00ff0000 0000ff00) bpp=32
> 2007-08-03 19:15:32.993 Typewriter[2803] got t=8
> 2007-08-03 19:15:32.993 Typewriter[2803] gamma=1.4
> 2007-08-03 19:15:32.997 Typewriter[2803] artcontext_setup_draw_info
> masks=(000000ff 0000ff00 00ff0000) bpp=32
> 2007-08-03 19:15:32.997 Typewriter[2803] got t=5
> 2007-08-03 19:15:32.997 Typewriter[2803] gamma=1.4
> 2007-08-03 19:15:33.048 Typewriter[2803] WARNING: Failed to retrieve
> MenuServer application proxy
> 2007-08-03 19:15:33.049 Typewriter[2803] WARNING: Failed to retrieve
> MenuServer application proxy
> 2007-08-03 19:15:33.050 Typewriter[2803] artcontext_setup_draw_info
> masks=(ff000000 00ff0000 0000ff00) bpp=32
> 2007-08-03 19:15:33.050 Typewriter[2803] got t=8
> 2007-08-03 19:15:33.050 Typewriter[2803] gamma=1.4
> 2007-08-03 19:15:33.054 Typewriter[2803] artcontext_setup_draw_info
> masks=(000000ff 0000ff00 00ff0000) bpp=32
> 2007-08-03 19:15:33.054 Typewriter[2803] got t=5
> 2007-08-03 19:15:33.054 Typewriter[2803] gamma=1.4
> 

Tkis output is a bit strange. The method artcontext_setup_draw_info,
which produces this log messages, is called only from ARTContext
initWithContextInfo:. This means that you have six initialisations of
art contexts, where I would expect to be only one. And these contexts
get alternately initialized for different byte orders. Sorry, I am
completely clueless. Could you try to get a backtrace from gdb for each
of these calls?
You could additionally try to disable the byte mask switching code in
ARTContext initWithContextInfo: and see how it affects your display.

Cheers,
Fred