Re: [Gnash-dev] FFmpeg parser issues

[Sandro Santilli]

On Fri, Jul 22, 2011 at 10:51:29PM +0200, Jonas 'Sortie' Termansen wrote:
> On 07/22/2011 10:43 PM, Sandro Santilli wrote:
> >
> >The misc-ming.all/NetStream-SquareTest.swf file should be loading
> >an .ogg file and the runner contains pixel-checking for it, so
> >it should be working, or we'd get a lot of failures on the
> >buildbots. Take a look.
> 
> The ogg file you mentioned does not have an audio track, and thus
> cleverly avoids the problem The problem is in AudioDecoderFfmpeg
> where it mistakenly requires a parsers and fails when none is
> available.
> 
> I have fixed that issue in my repo. I'll test things more properly
> and then I'll file a bug and patch when I am sure I fixed it
> correctly.

The most important thing would be an automated testcase.
That's of upmost importance to secure any bugfix.
For a start you may try to inject audio into the square.ogg file
and have the test runner (the .cpp file) check
for sound _not_ being silent. We could add an interface to
MovieTester to check that, which would then be reusable for
a testsuite focused on self-contained audio decoding tests.

--strk;

  ()   Free GIS & Flash consultant/developer
  /\   http://strk.keybit.net/services.html