[FFmpeg-devel] XVMC Deathmatch

[Diego Biurrun]

On Sat, Feb 14, 2009 at 05:13:38PM +0100, Michael Niedermayer wrote:
> On Sat, Feb 14, 2009 at 05:05:06PM +0100, Aurelien Jacobs wrote:
> > Michael Niedermayer wrote:
> > 
> > > >//the surface is needed for prediction, the codec manipulates this
> > > >#define MP_XVMC_STATE_PREDICTION 2
> > > >//this surface is needed for subpicture rendering
> > > >#define MP_XVMC_STATE_OSD_SOURCE 4
> > > >//                     1337    IDCT MCo
> > > >#define MP_XVMC_RENDER_MAGIC 0x1DC711C0
> > > >
> > > 
> > > 4 points
> > > replace MP prefixes by AV/FF depending on public/private and ensure that
> > > compatibility is provided until the next major ver bump but not beyond
> > 
> > Compatibility ??
> > This never was part of public API, so we shouldn't need to ensure
> > compatibility with anything...
> > (this applies to other part of the review)
> 
> mplayer/libvo/vo_xvmc.c:         p_render_surface_to_show->state |= MP_XVMC_STATE_OSD_SOURCE;
> mplayer/libvo/vo_xvmc.c:      if(src_rndr->state & MP_XVMC_STATE_OSD_SOURCE){
> mplayer/libvo/vo_xvmc.c:            src_rndr->state &= ~MP_XVMC_STATE_OSD_SOURCE;
> mplayer/libvo/vo_xvmc.c:                                  MP_XVMC_STATE_OSD_SOURCE |
> 
> well, it is used outside lavc so i thought i was supposed to be public ...

Used by anything apart from MPlayer?  I suggest we fix up MPlayer
afterwards and skip compatibility hacks.

Diego