[FFmpeg-devel] patch for DCA "floating point output" option

[Rich Felker]

On Sat, Jan 19, 2008 at 12:39:28PM -0500, Ronald S. Bultje wrote:
> Hi,
> 
> On Jan 19, 2008 12:38 PM, Rich Felker <dalias at aerifal.cx> wrote:
> 
> > On Sat, Jan 19, 2008 at 11:44:04AM -0500, Ronald S. Bultje wrote:
> > > On Jan 19, 2008 10:19 AM, madshi <dear at madshi.net> wrote:
> > > > Besides, I don't think that the name is useless. Currently the
> > decoders
> > > > all output "short" samples (shorter than they should be). "non short"
> > > > simply means that the decoders stop shorten the audio samples.
> > >
> > > So, video can "negotiate" an output (at least some decoders can), why
> > can't
> >
> > No, video decoders just always output the native format. Applications
> > using lavc must then convert if needed.
> 
> Sounds sane also. Let's do that for audio then (i.e. use enum SampleFormat
> for real!).

Well it's a lot less clear what's "native" for audio (i.e. more
dependent on the implementation - a fixedpoint implementation would
likely consider 16bit or 24bit integers to be native). For video it's
pretty clear.

Rich