[FFmpeg-devel] [PATCH v8 0/8] Properly decode ogg metadata in ogg/{vorbis, flac, opus} chained bitstreams

[Romain Beauxis]

Le jeu. 20 mars 2025 à 17:45, Michael Niedermayer
<michael at niedermayer.cc> a écrit :
>
> Hi Romain
>
> On Sun, Mar 16, 2025 at 07:59:59PM -0500, Romain Beauxis wrote:
> > Le dim. 16 mars 2025 à 19:20, Michael Niedermayer <michael at niedermayer.cc>
> > a écrit :
> > >
> > > On Sat, Mar 15, 2025 at 07:08:05PM -0500, Romain Beauxis wrote:
> > > > Hi all!
> > > >
> > > >
> > > > Le jeu. 13 mars 2025 à 08:43, Michael Niedermayer
> > > > <michael at niedermayer.cc> a écrit :
> > > > >
> > > > > Hi Romain
> > > > >
> > > > > On Tue, Mar 11, 2025 at 11:56:37AM -0500, Romain Beauxis wrote:
> > > > > > Le mar. 11 mars 2025 à 11:55, Romain Beauxis
> > > > > > <romain.beauxis at gmail.com> a écrit :
> > > > > > >
> > > > > > > This is a series of patches to allow proper decoding of ogg
> > metadata in chained
> > > > > > > `ogg/vorbis, `ogg/flac` and `ogg/opus` streams.
> > > > > > >
> > > > > > > ## Changes since last version:
> > > > > > > * Moved textual reference output for tests to test/ref/fate
> > > > > > > * Updated test binary to only output stream metadata when
> > > > > > >   AVSTREAM_EVENT_FLAG_METADATA_UPDATED flag is set.
> > > > > >
> > > > > > Again, forgot to add that test samples are available here:
> > > > > >
> > https://www.dropbox.com/scl/fo/xrtrna2rxr1j354hrtymq/AGwemlxHYecBLNmQ8Fsy--4?rlkey=lzilr4m9w4gfdqygoe172vvy8&dl=0
> > > >
> > > > Any update on this? Or concerns re: the samples?
> > > >
> > > > I've got more work ready to send, I'd love to finish this patch set to
> > > > get the next one rolling.
> > >
> > > the files contain "metadata: encoder=Lavc61.19.100"
> > >
> > > So they are version dependant which is a problem for reference files
> >
> > Can you elaborate on that?
> >
> > Vorbis comments standard requires that the encoder be the first comment in
> > the comment block and that's the value.
> >
> > The version written as the encoder is static metadata, I fail to see how it
> > would change with any future version of libavc, it is just a marker of the
> > version that was used when the file was created.
> >
> > If that can help, I'm happy to update the files to show any other value for
> > the encoder, though.
>
> If its just teh version read from a file, then its fine.
> The problem would be only if its the version of the code under test

Yes, this is only static text.

I tried with the -bitexact flag and the encoder is still created with
the current version.

Something that could also be patched later.

For now, I think that we can keep the current samples.

I'll post a new version of the patch series with the fix for
cross-compilation and also an update on pending_metadata to make it
use a av_tree in order to keep track of pts->metadata mappings in case
packets are not submitted in order.

-- Romain