On Mon, 8 Feb 2010, Mike Giacomelli wrote:

> The problem I see is that we do fixed point, while ffmpeg does floating
> point (except their MP3 decoder and lossless codecs).  Its not clear to me
> that they're all that interested in fixed point.

I just talked to them in #ffmpeg-devel. Them seem to be *very* interested in
getting our improvements, our codecs and in general eager to cooperate to get
things going. They also mentioned that their mp3 codec (ffmp3) is way faster
than libmad, which might be a reason for us to investigate how it behaves for
us in comparison.

> I think we owe Tremor a lot more then them.  Our version is something like
> 2x as fast as upstream at the moment on ARM.  And theres a lot of people out
> there using libTremor on ARM (sandisk, iRiver, Samsung , etc).  

I don't consider that any less reason to talk to ffmpeg, but just a reason to
also involve the Tremor people as well. Don't you?

To me, as a person not very closely involved in the codecs area, a way forward
could work like:

  - we setup a separate mailing list for fixed-point codec inter-project
    code sharing discussions
  - we invite Rockbox, ffmpeg and Tremor hackers (anyone else?)
  - we all speak up what we have and what we'd like from the others.

Then we work on making it all happen.

Opinions? Good? Bad?

-- 
  / daniel.haxx.se