[FFmpeg-devel] [PATCH] force dnxhd encoder to be independent of qsort internals
Sun Sep 20 12:09:07 CEST 2009
On Sun, Sep 20, 2009 at 01:52:21PM +0400, Michael Kostylev wrote:
> On Sun Sep 20 10:59:03 2009
> Reimar D?ffinger wrote:
> >>> this change makes sure there are no equal elements in the data passed to
> >>> qsort, thus making sure the result is independent of implementation
> >>> internals.
> >>> Not benchmarked, but compared to the function call overhead the extra
> >>> code in the comparison function should not hurt that much.
> >>> Regression test references need to be updated, too.
> >>> Tested and at least FreeBSD and Linux give the same results.
> >> Tested on OpenSolaris/i386, the encoding is ok, the decoding fails.
> > You are misreading the results, there is some issue with 1080i and dv,
> The dv test simply was always unstable. It may start to pass again after
> an unrelated commit to ffmpeg.c
I know, the point is that 720p so far looks to be stable. Also valgrind
does not detect any uninitialized data for me.
There is still the possibility that e.g. your compiler is broken.
Which is why I'd really like to get this patch applied ASAP so FATE can
help shake out any remaining issues.
More information about the ffmpeg-devel