[FFmpeg-devel] [FFmpeg-user] configure can't find existing x264 installationon Debian

Panagiotis Issaris takis.issaris
Fri Nov 14 13:50:46 CET 2008


Hi Benjamin,

On Fri, 2008-11-14 at 12:59 +0100, Benjamin Larsson wrote:
> M?ns Rullg?rd wrote:
> > Panagiotis Issaris wrote:
> >   
> >> Hi Baptiste,
> >>
> >> On Thu, 2008-11-13 at 19:41 -0800, Baptiste Coudurier wrote:
> >>     
> >>> Roman Shaposhnik wrote:
> >>>       
> >>>> On Nov 13, 2008, at 6:03 PM, Stephen Dewey wrote:
> >>>>         
> >>>>> Actually, just running svn update and recompiling seemed to take
> >>>>> care of
> >>>>> this issue for me. The problem seems to already be fixed in SVN,
> >>>>> although perhaps only coincidentally.
> >>>>> All problems solved! Thanks everybody!!
> >>>>>           
> >>>> Excellent!
> >>>>
> >>>> I do, however, have to question related to this issues:
> >>>>     1. shouldn't we finally make -cvslog list readonly so
> >>>>      that any comments get posted to -devel ? While
> >>>>      hunting down the commit that fixed the problem
> >>>>      I realized that Uoti has commented on the original
> >>>>      commit over at -cvslog. Of course, I overlooked his
> >>>>      email, since I don't actively track -cvslog.
> >>>>         
> >>> I find easier to answer to the related commit.
> >>>
> >>>       
> >>>>      2. Why don't we have any regression tests for H.264?
> >>>>      If it is a simple matter of adding some -- I can try to
> >>>>      take care of that tomorrow.
> >>>>         
> >>> Basically because we have no encoder, however we could put samples in
> >>> the test suite, it's in TODO AFAIK.
> >>>       
> >> Or we could add a basic encoder... ;-)
> >>     
> >
> > I am quite strongly opposed to adding basic (i.e. crap) versions of
> > anything, more so when a perfectly good external library exists and
> > is supported.  If a basic version exists, users will fail to enable
> > the proper support, and they'll get bad encoding results, ultimately
> > tarnishing the reputation of FFmpeg.  Furthermore, a trivial H.264,
> > if such a thing exists, will not exercise much of the decoder.
> >
> > I know you wrote a very basic H.264 encoder.  Don't take this personally.
> >
> >   
> 
> I on the other hand think we should add basic versions of encoders (like 
> mpeg2 audio encoder). But they should be disabled by default for the 
> same reasons as M?ns.

I fully agree. Initially, I couldn't understand why that commit (the one
that broke H.264 decoding) could have occurred and figured that
apparently no-one ran the testsuite... until I remembered that the
testsuite only tests decoding of files it has encoded.

That's not to say that this is sufficient of course, decoding should
also be tested on files encoded with other encoders or reference
bitstreams. But FATE takes care of that :-)


With friendly regards,
Takis






More information about the ffmpeg-devel mailing list