[Bf-committers] [13116] branches/blender2.5/blender/intern /ghost/intern: Add support for GHOST_kWindowState* to Linux.

[GSR]

Hi,
bdiego at gmail.com (2008-01-03 at 2204.16 +0100):
> 2) The Motif Hint isn't really a "FullScreen" mode, it's only a
> "NO DECORATIONS" options, so if the window manager don't support WM-spec

Why not call the function something like motifDecoration then? Or add
a comment in the function to remind the hack will be done later?

> this function only remove the border of the window, don't make it
> FullScreen. A simple fix (hack) is check if the Window Manager support
> WM-spec, in case that not, remove the decorations and move/resize the
> window.

OTOH, I guess WMs that do full screen a la video player when asked for
full screen is going to be a tricky case too.

Remember to update the command line params accordingly. Current ones
in trunk are wrong anyway, -w and "nothing" mean framed & maximized
with base size as big as screen, -W is full size but non maximized &
frameless, except when -p goes into play with different orders, -w is
affected but -W is not (if I did not get lost).

Now that I looked at the issue, the size hints are also wrong (trunk
and branch), the size fields are declared as user when I had not used
-p, they should be program instead. -p could move to standard X11 geom
param format, [<width>{xX}<height>][{+-}<xoffset>{+-}<yoffset>], the
code to implement that is freely avaliable, so no portability issues,
but more flexible and simpler param handling.
 
About the SGI code:

http://webcvs.freedesktop.org/mesa/Mesa/src/glut/glx/glut_init.c?content-type=text/plain&view=co

/* SGI optimization introduced in IRIX 6.3 to avoid X server
   round trips for interning common X atoms. */
#if defined(_SGI_EXTRA_PREDEFINES) && !defined(NO_FAST_ATOMS)
#include <X11/SGIFastAtom.h>
#else
#define XSGIFastInternAtom(dpy,string,fast_name,how) XInternAtom(dpy,string,how)
#endif

So you can try to figure all the fast names or better just use plain
X11 calls, as the request are not done constantly so the optimization
would be hardly noticeable.

GSR