[FFmpeg-devel] [PATCH 07/16] avformat/mxf: Deduplicate random_index_pack_key

Andreas Rheinhardt andreas.rheinhardt at gmail.com
Tue Jan 5 17:51:24 EET 2021


Tomas Härdin:
> mån 2021-01-04 klockan 01:28 +0100 skrev Andreas Rheinhardt:
>> Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt at gmail.com>
>> ---
>> There is also another key that is duplicated: header_open_partition_key
>> and mxf_jp2k_rsiz. But they seem to mean something different which makes
>> me wonder if one of these keys is incorrect; I have therefore refrained
>> from deduplicating them.
> 
> Looks like I didn't notice this when I reviewed the patch. This
> indicates that the MXF muxer that wrote the file that 2af260e3 adds
> support for is broken.
> 
> S422m-2006 says that the UL designator for Rsiz should be
> 04.01.06.03.01.00.00.00. S400m-2004 defines the prefix to this
> designator, which is the usual 06.0E.2B.34.04.01.01.VV where VV is the
> registry version.
> 
> Did we have a sample for this?
> 
I don't have any sample for it; it is also untested by FATE.

> If you decide to deduplicate rsiz then you should have kind of comment
> to this effect, and/or a link to this thread and 
> http://ffmpeg.org/pipermail/ffmpeg-devel/2015-March/169591.html
> 
If the usage of this key in the demuxer is wrong, then this should be
fixed and not deduplicated.

> The patch itself looks fine. Might as well deduplicate all keys.
> There's little sense in having separate sets of keys defined for the
> muxer and demuxer.

The muxer has some of its ULs in a structure that contains pointers to
muxing-related functions, so that can't be deduplicated without lots of
modifications. And then there are several d-10 keys only used by the
muxer. Is this intended?

- Andreas


More information about the ffmpeg-devel mailing list