On 06/03/2010 11:13 PM, Reimar D?ffinger wrote: > On Thu, Jun 03, 2010 at 10:15:17PM +0200, Alexander Roalter wrote: >> On 06/03/2010 09:20 PM, Reimar D?ffinger wrote: >>> On Thu, Jun 03, 2010 at 02:20:06PM +0200, Alexander Roalter wrote: >>>> On 06/03/2010 01:32 PM, Alexander Roalter wrote: >>>>> my last update was r31257, updated to r31309 today and -xineramascreen >>>>> doesn't work anymore. regardless of the number I enter, the window opens >>>>> on the screen the mouse cursor is at. >>>>> >>>>> >>>> The regression was introduced before r31289, probably by >>>> >>>> r31289 | reimar | 2010-05-30 21:14:18 +0200 (Sun, 30 May 2010) | 2 lines >>>> Change code to let Window Manager chose window location by default. >>> >>> For which the Changelog says: >>> >>> * X11: Window manager choses Window position by default. >>> Add geometry=50%:50% to your configuration to get the old behaviour. >>> >>> So unless this does not help, or MPlayer does not go to fullscreen on the >>> intended screen I don't see any regression. >>> If you know of a window manager hint to have it do automatic placement but >>> preferring a certain screen tell me. >> >> geometry 50%:50% does work as expected, but I'd have a different >> suggestion (don't know how this fits into it at all:) >> >> setting xineramascreen to screen n (n >= 0), as there's no way to tell >> the window manager to do automatic placement on a certain screen, should >> automatically have the old behavior, even if it means windows are >> positioned in the center of the screen. > > Well, I just realize that's what that chunk of code on my home machine > (in a different country as I am unfortunately) was supposed to do... > I just forgot to decide which is preferable, since after all xineramascreen > is not useless without this, it still determines where it goes for fullscreen. > >> xineramascreen=-1 has now the same effect as if no xineramascreen is >> given, b/c the screen is automatically positioned where the cursor is > > Huh? I can't see how this makes any difference for xineramascreen == -1? > This is my patch restored from memory, please test: > > Index: libvo/video_out.c > =================================================================== > --- libvo/video_out.c (revision 31310) > +++ libvo/video_out.c (working copy) > @@ -356,6 +356,7 @@ > vo_dy = (int)(vo_screenheight - d_height) / 2; > geometry(&vo_dx, &vo_dy, &d_width, &d_height, > vo_screenwidth, vo_screenheight); > + geometry_xy_changed |= xinerama_screen >= 0; > vo_dx += xinerama_x; > vo_dy += xinerama_y; > vo_dwidth = d_width; With this change, it behaves as expected: -xineramascreen 0 plays on the 1st screen, -xineramascreen 1 on the 2nd, -xineramascreen 2 on the 3rd, and -xineramascreen -1 plays on the current screen (with window manager placement, not in the center) I'm happy with the behavior such as it is now. I don't see the connection with xinerama and fullscreen, though: pressing f during playback switches to fullscreen on the screen the window (or at least the greater part of it) is in... -- Cheers, Alex