Re: [FFmpeg-devel] [PATCH] lavc/libx265: flag as experimental

[Anton Khirnov <anton@khirnov.net>]

Quoting Tobias Rapp (2023-03-15 16:00:04)
> On 15/03/2023 15:45, Anton Khirnov wrote:
> 
> > This encoder leaks and overreads, as can be seen e.g. by running an
> > encode under valgrind with default encoder parameters. This was known
> > upstream since at least 2019 (e.g. bitbucket issue #482) but never fixed
> > until now.
> >
> > Since upstream does not seem to practice basic code hygiene, make sure
> > people do not use this encoder without knowing what they are getting
> > into.
> > ---
> >   libavcodec/libx265.c | 3 ++-
> >   1 file changed, 2 insertions(+), 1 deletion(-)
> >
> > diff --git a/libavcodec/libx265.c b/libavcodec/libx265.c
> > index 420d0953af..d4511251a5 100644
> > --- a/libavcodec/libx265.c
> > +++ b/libavcodec/libx265.c
> > @@ -895,7 +895,8 @@ FFCodec ff_libx265_encoder = {
> >       .p.id             = AV_CODEC_ID_HEVC,
> >       .p.capabilities   = AV_CODEC_CAP_DR1 | AV_CODEC_CAP_DELAY |
> >                           AV_CODEC_CAP_OTHER_THREADS |
> > -                        AV_CODEC_CAP_ENCODER_REORDERED_OPAQUE,
> > +                        AV_CODEC_CAP_ENCODER_REORDERED_OPAQUE |
> > +                        AV_CODEC_CAP_EXPERIMENTAL,
> >       .p.priv_class     = &class,
> >       .p.wrapper_name   = "libx265",
> >       .init             = libx265_encode_init,
> 
> My understanding of this flag for encoders is that it indicates some 
> possible experimental/non-standard bitstream features contained in 
> output data, not that the implementation itself is considered sub-par.

Our internal aac encoder used to be marked as experimental. vorbis and
opus still are.
So it's clearly about the implementation quality rather than standard
compliance.

-- 
Anton Khirnov
_______________________________________________
ffmpeg-devel mailing list
ffmpeg-devel@ffmpeg.org
https://ffmpeg.org/mailman/listinfo/ffmpeg-devel

To unsubscribe, visit link above, or email
ffmpeg-devel-request@ffmpeg.org with subject "unsubscribe".