[FFmpeg-devel] Error regression test

[Frans de Boer]

On Tue, 2009-06-02 at 19:08 -0300, Ramiro Polla wrote:
> On Tue, Jun 2, 2009 at 6:33 PM, Frans de Boer <frans at fransdb.nl> wrote:
> > On Tue, 2009-06-02 at 22:20 +0100, M?ns Rullg?rd wrote:
> >> Ramiro Polla <ramiro.polla at gmail.com> writes:
> >> > On Tue, Jun 2, 2009 at 4:13 PM, Frans de Boer <frans at fransdb.nl> wrote:
> >> >> When running the tests, I get at some point an error.
> >> > [...]
> >> >> Any suggestions?
> >> >
> >> > Yes. Read http://ffmpeg.org/bugreports.html and file a more complete
> >> > bug report on the roundup bug tracker.
> >>
> >> One more: don't hijack threads.
> >>
> >> Also, don't expect any help here, given your history of trolling.
> >
> > First, I am not aware of hijacking a thread. To me it was a new thread.
> 
> Please look at:
> http://lists.mplayerhq.hu/pipermail/ffmpeg-devel/2009-June/thread.html#start
> 
> Your message spawned from the "[FFmpeg-devel] [PATCH]
> avcodec_decode_video must set PKT_FLAG_KEY" thread. That is called
> thread hijacking. Google can help you identify why that happened and
> how to avoid it.
> 
> [...]
> > For the other people here: The bug report page mentioned the ffmpeg
> > itself. If I do 'gdb make test', that will work too?
> 
> The bugreports page says to use gdb "If you encounter a crash bug",
> which is not your case.
> 
> make test works by generating some sample image/audio, encoding it
> with a bunch of encoders, and decoding that. It will then check that
> the md5sum of the encoded and decoded files match a reference which is
> in the SVN repo.
> 
> The output from make test is not supposed to be a bug report "as-is",
> but it should help you locate a bug. You should try to find out why
> the encoded/decoded files differ on your platform (preferably by
> comparing against another system where make test passes).
> 
> Ramiro Polla

Thanks for your reaction. I assumed wrongly that changing the subject
line would start a new thread. I offer my apologies.

I will also try to follow your suggestions and will post an update -
starting a proper thread - if and when I have found the offending
code/data.

Frans.