[FFmpeg-devel] [libav-devel] [PATCH 1/6] avutil: remove deprecation of qscale_{table, type}, qstride

Hendrik Leppkes h.leppkes at gmail.com
Sat Aug 8 17:58:01 CEST 2015


On Sat, Aug 8, 2015 at 2:49 PM, wm4 <nfxjfg at googlemail.com> wrote:
> On Sat, 8 Aug 2015 14:31:21 +0200
> Hendrik Leppkes <h.leppkes at gmail.com> wrote:
>
>> On Sat, Aug 8, 2015 at 1:36 PM, Andreas Cadhalpun
>> <andreas.cadhalpun at googlemail.com> wrote:
>> > They are used by the not deprecated av_frame_{g,s}et_qp_table.
>> >
>> > Signed-off-by: Andreas Cadhalpun <Andreas.Cadhalpun at googlemail.com>
>> > ---
>> >  libavutil/frame.h | 6 ++----
>> >  1 file changed, 2 insertions(+), 4 deletions(-)
>> >
>> > diff --git a/libavutil/frame.h b/libavutil/frame.h
>> > index 196b578..c4e333c 100644
>> > --- a/libavutil/frame.h
>> > +++ b/libavutil/frame.h
>> > @@ -285,21 +285,19 @@ typedef struct AVFrame {
>> >  #if FF_API_AVFRAME_LAVC
>> >      attribute_deprecated
>> >      int reference;
>> > -
>> > +#endif
>
> Stray change.
>
>> >      /**
>> >       * QP table
>> >       */
>> > -    attribute_deprecated
>> >      int8_t *qscale_table;
>> >      /**
>> >       * QP store stride
>> >       */
>> > -    attribute_deprecated
>> >      int qstride;
>> >
>> > -    attribute_deprecated
>> >      int qscale_type;
>> >
>> > +#if FF_API_AVFRAME_LAVC
>> >      /**
>> >       * mbskip_table[mb]>=1 if MB didn't change
>> >       * stride= mb_width = (width+15)>>4
>>
>>
>> Didn't this stuff move into sidedata
>
> In FFmpeg. It's completely gone in Libav. (FFmpeg "needs" it for their
> relatively useless postproc filters.)
>
> Removing the deprecation won't make this work either; it just makes
> projects referencing it compile. And apparently distros can't be
> bothered to patch this, even though making sure the projects actually
> _work_ as opposed to merely compiling them got to be much more work.
> Makes no sense to me.

Then we should move it into side data, just like all the other video
metadata which was already moved, it has no place in a generic AVFrame
(and makes it consistent with the other stuff as well).

- Hendrik


More information about the ffmpeg-devel mailing list