[Bf-committers] Re: CVS commit: blender/intern/ghost/intern GHOST_WindowX11.cpp

[GSR]

Hi,
cbarton at metavr.com (2007-04-15 at 0802.12 +0200):
> campbellbarton (Campbell Barton) 2007/04/15 08:02:12 CEST
> 
>   Modified files:
>     blender/intern/ghost/intern GHOST_WindowX11.cpp 
>   
>   Log:
>   last commit didnt include this file.
>   
>   Revision  Changes    Path
>   1.19      +24 -4     blender/intern/ghost/intern/GHOST_WindowX11.cpp
>     <http://projects.blender.org/viewcvs/viewcvs.cgi/blender/intern/ghost/intern/GHOST_WindowX11.cpp.diff?r1=1.18&r2=1.19&cvsroot=bf-blender>

It was a patch for review, more than direct commit. The idea of
checking all atoms seems sane, as in your patch. And anyway, it does
not behave as expected (yet? is it ever fixable? what is to be
expected?).

Try blender, blender -w and blender -W (and use xprops to check each
case, do not just believe what the WM forces and shows you). The tests
should also try with -p, before and after the -w/-W, but I am not
going to test and patch those without knowing what is the desired
behaviour (but -p varies things, that is sure). In my machine I see:

            Option   none       -w         -W
Start size            FP         OM         FM
Start state           Maximized  Normal     Normal
First unmaximize      OM         n/a        n/a
First maximize        n/a        FP         FP
Maximized state hint  Y          N          N
Panels avoided        Y          N          N
Motif decor hint      N          N          Y
Frames                Y          Y          N

FP Fills all area up to panels
OM Overflows monitor (as increased by frame decors)
FM Fills monitor (ignoring panels)

You can blame all you want the wm about some size decisions, but not
the hints, and none != -w (as -h implies). And the requested size is
always monitor size (until you add -p to the mix).

GSR