[MPlayer-dev-eng] [PATCH] Xinerama full screen bug fix

Georgi Chorbadzhiyski gf at unixsol.org
Thu Jul 24 17:48:59 CEST 2008


Around 07/24/08 18:22, Reimar Döffinger scribbled:
> Hello,
> On Wed, Jul 23, 2008 at 01:52:54AM +0300, Georgi Chorbadzhiyski wrote:
>> The following patch fixes a problem with full screen geometry when
>> mplayer is started with the X11 -xineramascreen option.
>>
>> Without this patch the following produces wrong results:
>>   1. Start mplayer -xineramascreen 1 file
>>   2. Hit 'f' to put mplayer in full screen mode (or use -fs option)
>>      - Everything works as expected
>>   3. Hit 'f' again to stop full screen mode
>>   4. Move mplayer window to another screen (with different resolution)
>>   5. Hit 'f' to put mplayer into full screen mode again
>>     - Expected - mplayer in full screen mode
> 
> "on the monitor specified by -xineramascreen". Which it will do if
> you do as the man page suggests:
> 'Will usually only work with "\-fstype \-fullscreen" or "\-fstype
> none".'
> 
>> The function auto detects current screen and acts accordingly
>> when no -xineramascreen option is used, but fails to detect current
>> screen when -xinramascreen is used.
> 
> It is not supposed to.
> 
>> The proposed fix is simple, allow vo_x11_fullscreen() to ignore
>> xinerama_screen option and always force auto detection in case when
>> full screen is requested.
> 
> It is also wrong, it breaks the (somewhat) consistent behaviour that
> -xineramascreen finally has.
> I tried to improve it but the xineramascreen description in the man page
> is still not good.
> xineramascreen currently sets 3 things: the screen on which MPlayer
> starts (unless the window manager interferes), the screen on which
> MPlayer goes to fullscreen mode (unless the window manager interferes,
> in which case the behaviour is as you describe) and the screen relative

I'm using kwin as WM so should I fill a bug about it for this behavior?

> to which -geometry works.
> Your particular patch would at least break consistency with behaviour on
> Windows, and any patch that does not add a new option would at least remove
> a currently working feature (going to fullscreen always on a certain
> screen, no matter where the window is).

Ok, so what is the solution so the sequence described above works
as I expected (go to full screen on which the mplayer window is)?

Probably using -geometry? If this is the recommended option should
I send patch to manual page, describing xineramascreen option behavior
wrt full screen and mplayer window position?

Thanks for the reply.

-- 
Georgi Chorbadzhiyski
http://georgi.unixsol.org/



More information about the MPlayer-dev-eng mailing list