[FFmpeg-devel] [PATCH] fftools/ffprobe: fix printing side data components and pieces

[Stefano Sabatini]

On date Tuesday 2024-03-05 11:38:20 -0300, James Almer wrote:
> On 3/5/2024 11:34 AM, Stefano Sabatini wrote:
> > On date Monday 2024-03-04 15:56:46 -0300, James Almer wrote:
> > > Components and pieces are side data specific fields and there's a variable
> > > amount of them.
> > > They also need to be identified in some form, so print a type too.
> > > 
> > > Signed-off-by: James Almer <jamrial at gmail.com>
> > > ---
> > > You can see the difference with
> > > 
> > > ffprobe -show_frames -of compact $fate_samples/mov/dovi-p7.mp4
> > > ffprobe -show_frames -of xml $fate_samples/mov/dovi-p7.mp4
> > > 
> > >   fftools/ffprobe.c | 33 +++++++++++++++++++++++++++------
> > >   1 file changed, 27 insertions(+), 6 deletions(-)
> > > 
[...]
> > LGTM. Not clear if this is fixing a regression or if the format (at
> > least for XML/compact) was always wrong.
> 

> I'm on the fence on this. Try removing the SECTION_FLAG_HAS_TYPE from
> SECTION_ID_FRAME_SIDE_DATA_COMPONENT and SECTION_ID_FRAME_SIDE_DATA_PIECE
> and the output is broken again for the compact output. So at least
> compact_print_section_header() may be faulty. I did not investigate much
> because I'm not familiar with the code.

That is, if you define the type (as in the current implementation) it
looks correct?

> In any case, this change is needed because of the
> SECTION_FLAG_HAS_VARIABLE_FIELDS addition for side data where the printed
> fields are not a fixed amount.

Agree.