On Fri, Jan 11, 2013 at 10:24:49PM +0200, Rémi Denis-Courmont wrote:
> Le vendredi 11 janvier 2013 22:01:13, Diego Biurrun a écrit :
> > On Fri, Jan 11, 2013 at 08:33:20PM +0200, Rémi Denis-Courmont wrote:
> > > --- a/libavcodec/vdpau.c
> > > +++ b/libavcodec/vdpau.c
> > > @@ -38,6 +38,55 @@
> > > +/* Obsolete non-hwaccel VDPAU support below... */
> > > +
> > >
> > > void ff_vdpau_h264_set_reference_frames(MpegEncContext *s)
> > > {
> >
> > Should that obsolete support be dropped at the next version bump?
>
> Is that a question to me? I can provide a separate patch later, but it would
> break mplayer's contemporary VDPAU support.
No, it's a question to everyone. If at all it should be properly deprecated
by FF_API_FOO ifdefs that evaluate to 0 at the next major bump.
> > > --- a/libavcodec/vdpau_internal.h
> > > +++ b/libavcodec/vdpau_internal.h
> > > @@ -25,8 +25,23 @@
> > >
> > > +/** Extract VdpVideoSurface from a Picture */
> > > +static inline VdpVideoSurface ff_vdpau_get_surface_id(Picture *pic)
> > > +{
> > > + return (uintptr_t)pic->f.data[3];
> > > +}
> >
> > uintptr_t vs. VdpVideoSurface - I'm confused ...
>
> GCC will bitch if casting a pointer to an integer of smaller entropy. Note
> that VdpVideoSurface is an unsigned 32-bits descriptor, as is typical of X11
> resources.
>
> Do you really want this instead?
>
> return (VdpVideoSurface)(uintptr_t)pic->f.data[3];
No.
Diego
_______________________________________________
libav-devel mailing list
[email protected]
https://lists.libav.org/mailman/listinfo/libav-devel
