[FFmpeg-devel] [PATCH 2/2] avformat/id3v2: Check that decode_str() did advance

[softworkz .]

> -----Original Message-----
> From: ffmpeg-devel <ffmpeg-devel-bounces at ffmpeg.org> On Behalf Of
> Michael Niedermayer
> Sent: Mittwoch, 16. April 2025 03:22
> To: FFmpeg development discussions and patches <ffmpeg-
> devel at ffmpeg.org>
> Subject: Re: [FFmpeg-devel] [PATCH 2/2] avformat/id3v2: Check that
> decode_str() did advance
> 
> On Tue, Apr 15, 2025 at 10:59:07PM +0000, softworkz . wrote:
> >
> >
> > > -----Original Message-----
> > > From: ffmpeg-devel <ffmpeg-devel-bounces at ffmpeg.org> On Behalf Of
> > > Michael Niedermayer
> > > Sent: Mittwoch, 16. April 2025 00:50
> > > To: FFmpeg development discussions and patches <ffmpeg-
> > > devel at ffmpeg.org>
> > > Subject: Re: [FFmpeg-devel] [PATCH 2/2] avformat/id3v2: Check that
> > > decode_str() did advance
> > >
> > > On Tue, Apr 15, 2025 at 07:59:00PM +0000, softworkz . wrote:
> > > [...]
> > >
> > > > - The representation of multi-values - both, internally and when
> > > >   outputting as probe data - is a de-facto standard
> > >
> > > The external handling in formats is specified in the corresponing
> > > specifications. ";" is certainly not correct for formats which
> > > natively support multiple values per key.
> > >
> > > Internally, if you have a data structure that represents multiple
> > > authors, you certainly do not set it to one author and a string
> > > with a bunch of semicolons seperating multiple authors
> > >
> > > Title: "Smile ;)"
> > > Author "Smily Face ;)"
> > >
> > > is not 2 Titles and not 2 Authors and software that cannot handle
> that
> > > should not be used as reference IMHO
> > >
> > > That said, anything that works is fine with me,
> > >
> > > But internally it will be better to use a representation that is
> > > universal, generic and simple, ";" may seem to be that but only
> > > as long as you do nothing with it and dont care about corner cases
> > >
> > > Ill leave this ";" question to everyone else, i have a backlog
> > > of quite a few things i need to work on
> >
> > This is not a great outcome, because "leaving everyone else" means
> > nothing will happen.
> 
> I have a release to do, I have contracts that i should be working on,
> as iam a few month behind some deadline, I have a backlog of fuzzer
> issues i wanted to look at. I have a backlog of other things
> 
> And this ";" thing is close to the last thing i want to spend my time
> on ;)

It's fine. You don't have to. I'm also not insisting on my patch.


> > At least revert the rcombs patch until there's a conclusion, because
> > it really makes things worse than better with regards to FFprobe
> > output.
> 
> I guess thats reasonable.

Perfect, thanks.


> > This will cause deserialization errors for many people in the world
> > who are processing FFprobe data.
> 
> As said, ffprobe should not produce troublesome output

As I said, it cannot be remedied on the FFprobe side without making a 
breaking change to its output format, that's what turns this small
thing into a big one.

Thank you
sw