[MPlayer-dev-eng] [PATCH] honour screenw/screenh with xinerama on

[Olivier Jolly]

Hi,

Reimar Döffinger a écrit :
> While it makes the vo_winvidix and vo_ggi cases a bit more complex, I
> think it makes more sense to change cfg-mplayer.h to use the vo_user_
> stuff.
>   
I have the feeling that we need to have 2 sets of screenw/h values. One 
which is the user will, constant after the command line parsing and one 
other which is the working set (and can be updated by autodetection 
including xinerama or directx related functions). If you're ok with me 
on this principle, I
can understand that we could fill in the vo_user_* with a configuration 
in cfg-mplayer.h, ok. But still, we need to prefill the working set 
(vo_*) with those vo_user_* counterpart before calling any config 
methods. This makes an even number of move (but maybe slightly easier to 
understand).
What I don't understand is why that would make vo_winvidix and vo_ggi 
handling different. They would be feeded with the same content of (vo_*) 
and with the ability to restore values (or to avoid changing them 
actually, if we're sure they have the right value before entering any 
block overriding the vo_* values).
I am probably missing something there.
> Like this you can get rid of the two "// Keep command line values for
> screen height and width" things (esp. since you would actually need a
> third one for your patch to vo_directx to work, since that does not use
> w32_common.c at all :-(
>   
oh, that's another story. I think I used the last switch option of 
control in vo_directx to override the fullscreen size detection by 
vo_user_* stuff but since I don't have any windows machine, I couldn't 
test this. But using the cfg-mplayer.h to feed vo_user_*, those values 
would be correctly fed, ok; anyway wouldn't we need to set the value of 
vo_screenwidth/height at some point ? Which would be expected in a 
vo_init like function. There is no such thing for directx ?

Thanks for your reply
> Greetings,
>   
Take care
> Reimar Döffinger
+Olivier