[FFmpeg-devel] yesterdays libavfilter merge

[Carl Eugen Hoyos]

Nicolas George <nicolas.george <at> normalesup.org> writes:

> A. there are bugs fixed in ffmpeg and not libav;

> B. there are bugs fixed in libav but not in ffmpeg due to 
>    different code paths;
> C. there are bugs in ffmpeg introduced by the daily merge;
> D. there are bugs in ffmpeg introduced by our own changes.
> 
> And while B and C are quite infrequent, I would be surprised 
> if B+C+D did not, to a certain amount, compensate A.

If you divide A by 10, maybe.

> I am perfectly ready to accept your claim if you are sure of it.

I stopped counting when the number of regressions reached ~100, 
that was many months ago, and the rate did not change 
significantly since.

> A few concrete examples would be nice, though, to serve 
> as arguments in a discussion

That is extremely simple imo:
Just look (as in: test) at the open bugs in bugzilla (and roundup) 
and the closed tickets on trac. If you don't trust me, you could 
of course look (=test) at the closed bugs in bugzilla and the open 
ones in trac, but this will not show much difference.
(I am of course not counting bugs/missing features in features 
that do not exist in the qatar fork, but even if you do, it will 
not strongly change the result.)

> for example if someone should decide to take the matter to the
> Debian community.

Debian admins have repeatedly stated that they do neither care 
about bugs nor about regressions.

Carl Eugen