[MPlayer-users] Unable to play aac stream
Reimar.Doeffinger at stud.uni-karlsruhe.de
Sat Apr 5 23:32:16 CEST 2008
On Sun, Apr 06, 2008 at 06:58:05AM +1000, Ross Gayler wrote:
> > Fixed in SVN r26331, r26301 broke audio for almost all (admittedly the
> more obscure ones only) formats...
> How do the people who work on MPlayer do regression testing?
> If many of the issues that get fixed in MPlayer development are to do with
> working around dumb things that content-providers have done, then the
> MPlayer regression test library would have to contain lots of simulators of
> broken content streams. I don't know anything about the technicalities of
> this, but making a library of simulators of broken streams sounds much more
> difficult than the effort required to write MPlayer itself.
Yes, which is why we use the most advanced things available to us: You,
the users ;-). The unfortunate truth is we do not have regression tests
apart from that, though we have a huge library of samples a
samples.mplayerhq.hu to check things.
In addition, we are in the process of moving the most critical stuff
(demuxers, decoders, in the future also filters and possibly streams) to
FFmpeg, which does have true regression tests, both in the form of "make
test" and also at http://fate.multimedia.cx/ (let's do a little
stress-test on that server ;-) ).
Though that still leaves things like audio and video output in MPlayer,
and for most of these I do not even have an idea how one could do any
kind of regression test.
More information about the MPlayer-users