Opened 12 years ago

Last modified 8 years ago

#861 assigned defect

screenw/screenh is ignored in XV with xinerama on

Reported by: olivier.jolly@… Owned by: reimar
Priority: normal Component: vo
Version: HEAD Severity: normal
Keywords: Cc: reimar
Blocked By: Blocking:
Reproduced by developer: Analyzed by developer:

Description

When mplayer is compiled with xinearama support, the update_xinerama_info function will override the vo_screenheight and vo_screenwidth inconditionnaly.
When using the -screenw / -screenh to map a part of the screen to TVout (using the nvtv tool for example), since 1.0pre8, the whole screen was used for displaying.
I'm proposing a patch in the next comment.

Attachments (1)

xinerama_override_screenwh.patch (2.2 KB) - added by olivier.jolly@… 12 years ago.
Patch proposition to honour screenw/screenh even in xinerama

Download all attachments as: .zip

Change History (6)

Changed 12 years ago by olivier.jolly@…

Patch proposition to honour screenw/screenh even in xinerama

comment:1 Changed 12 years ago by olivier.jolly@…

In this patch, I'm keeping a value to know whether the vo_screenheight/width are set by the default values (and are flagged as overridable by the xinerama_update_info function) or not (and can't be overrided).
I think the indentation isn't right in the patch even if I took care of it in emacs (with spaces to keep the same indenetation as existing code and such).

comment:2 Changed 12 years ago by beastd

  • Status changed from new to assigned

(In reply to comment #1)

Created an attachment (id=378) [details]
Patch proposition to honour screenw/screenh even in xinerama

In this patch, I'm keeping a value to know whether the vo_screenheight/width
are set by the default values (and are flagged as overridable by the
xinerama_update_info function) or not (and can't be overrided).

I looked at the patch and my first impression is it seems to be quite OK.

Am i correct that you tested it and it works for your problem and otherwise
the behaviour stays the same as before?

Other thing is that i tend to keep the changes only to the X11 code as it

doesn't apply to other video outputs modules. Perhaps it could but then it
would be better the flag would be set elsewhere in the vo layer code.

I think the indentation isn't right in the patch even if I took care of it in
emacs (with spaces to keep the same indenetation as existing code and such).

Will have a look at the indentation when testing the patch (hopefully the

next days).

comment:3 Changed 12 years ago by olivier.jolly@…

(In reply to comment #2)

Am i correct that you tested it and it works for your problem and otherwise
the behaviour stays the same as before?

Right, I made some tests and it was ok.

Other thing is that i tend to keep the changes only to the X11 code as it

doesn't apply to other video outputs modules. Perhaps it could but then it
would be better the flag would be set elsewhere in the vo layer code.

Indeed, since it's xinerama linked and that it is x11 specific, it might be better to keep it elsewhere. Can you point me out where I should fit those variables to be cleaner ?

comment:4 Changed 12 years ago by reimar

  • Cc Reimar.Doeffinger@… added

I don't think this solution is any good, it is only a hack. The same problem exist also for -vo directx and -vo gl under windows and possibly in a few more cases as well.
I did not yet have the time and interest to think up a proper solution, but probably the user-selected width and height should be saved in some other variable so instead of having to check for it all over the init code you can just "restore" them after init.

comment:5 Changed 8 years ago by compn

  • Owner changed from beastd to reimar
Note: See TracTickets for help on using tickets.