FreeLists / openbeosmediakit / [openbeosmediakit] Re: 3ivx decoder and Haiku

["Axel Dörfler" <axeld@xxxxxxxxxxxxxxxx>]

"Andrew Bachmann" <shatty@xxxxxxxxxxxxx> wrote:
> Marcus Overhagen <marcusoverhagen@xxxxxxxx> wrote:
> > > As an initial hint, why not? As a codec writer, I should have a 
> > > good 
> > > idea about how good my codec is at parsing a certain stream?!
> > It adds complexity. Each codec would have to publish for each 
> > format
> > a confidence or priority level. I don't think that adding all this 
> > does bode 
> > well.
> I agree with Marcus.  In fact, I would go farther and say that the 
> "right"
> way to handle codec capability reporting would have to supply the
> stream itself to the codec.  It's unreasonable to expect that the 
> extractor
> would be able to fully characterize the stream in all possibly 
> relevant
> ways.  This would require some restructuring and is not necessary for 
> R1.

Yes, I think this makes some sense. But the codec should only be able 
to deny the request then; it shouldn't be necessary to ask all codecs 
for proper support in that case.

Bye,
   Axel.