[MPlayer-dev-eng] TODO for release? bugs status...

Jindrich Makovicka makovick at kmlinux.fjfi.cvut.cz
Wed Jan 29 19:01:58 CET 2003


Arpi wrote:

>Hi,
>
>  
>
>>Here is another iteration which seems to solve both issues.
>>    
>>
>
>sure? :)
>
>i've tried to play 2 avi files, none of them have prescaling (aspect ratio
>coded in file) but the first is 16:9 the second is 4:3 ratio.
>
>mplayer file1.avi file2.avi
>worx
>
>mplayer -fixed-vo file1.avi file2.avi
>worx, but the second file has wrong (16:9) aspect ratio.
>
aaargh, it's the **** fs switching code again !
this is definitely post-0.90 issue

>mplayer -fixed-vo file1.avi file2.avi
>while playing file1, press 'f' and then set panscan to 100%
>file2 will result pure black screen.
>
i forgot that when clearing the window, it doesn't have to repaint 
itself necessarily. fixed. (?)

>now try the same, increase panscan but don't set panscan to 100%
>file1 & file2 plays ok.
>
>also, your change to saving panscan may break other vo drivers supporting
>panscan (xmga etc) even without -fixed-vo :(
>(-> bad idea in 0.90-rc times)
>
sounds reasonable. anyway we won't get rid of such problems until 
someone rewrites the whole vo layer.

>i vote for applying your first(!) patch, it fixed -fixed-vo
>or applying second patch with fix for panscan.
>this vo_panscan_start thing is not acceptable now :(
>
ok, now i see i should have more sleep instead of coding tonight ;)

>or just forget the whole thing. -fixed-vo is still experimental feature.
>and after the release someone should go and re-design this whole -fs vs 'f'
>vs panscan etc thingie.
>all these flags should be changed to 3-state: yes, no, previous.
>so user could override it (both force enable and disable) from commandline,
>but by default it would preserve 'f', panscan etc states.
>another problem is that vo_fs is actually an _internal_ variable of
>x11_common's fs switching code, it shouldn't be read/write by vo drivers.
>or we should change vo_x11_fullscreen to use vo_fs as parameter instead of
>handling fs toggle itself. (so 'f' handler could do vo_fs=!vo_fs; vo_x11_fullscreen())
>
yes, we wouldn't need then the ugly hacks detecting whether it's 
actually fullscreen or not.

>and at the same time remove the fullscreen flag from config()'s last
>parameter or 'emulate' it by vo_fs for compatibility...
>  
>
patch changing only vo_xv with the black screen fix is attached.

-- 
Jindrich Makovicka

-------------- next part --------------
An embedded and charset-unspecified text was scrubbed...
Name: vo_xv.c.diff7
URL: <http://lists.mplayerhq.hu/pipermail/mplayer-dev-eng/attachments/20030129/462ab2c0/attachment.asc>


More information about the MPlayer-dev-eng mailing list