[FFmpeg-devel] [PATCH] Add af_resample - sample fmt and channel layout conversion filter
Mon Sep 27 13:44:30 CEST 2010
On Mon, Sep 27, 2010 at 12:22:49AM -0700, S.N. Hemanth Meenakshisundaram wrote:
> On 09/26/2010 05:50 PM, Michael Niedermayer wrote:
> > On Mon, Sep 27, 2010 at 02:41:16AM +0200, Stefano Sabatini wrote:
> >> On date Monday 2010-09-27 02:21:49 +0200, Michael Niedermayer encoded:
> >>> On Mon, Sep 27, 2010 at 12:40:31AM +0200, Stefano Sabatini wrote:
> >> [...]
> >>>>> the functions belong to libavfilter
> >>>>> what is the plan to move them there?
> >>>> Impossible as lavc uses that functions internally. The plan was to
> >>> resample uses them, resample doesnt belong to lavc, not more than
> >>> sws would belong there
> >>>> directly use them, and move to something better when we'll have a
> >>>> revisited resampling/conversion API, a libavresample was mentioned.
> >>>> Since this is not going to take small time, the idea was to use the
> >>>> quickest solution and not stall audio lavfi development.
> >>> the quickest is to copy the 2 pages of poor code not to export it as
> >>> public ABI/API from the wrong lib
> >>> that API/ABI requires maintaince and is not a small burden when things
> >>> have to be moved to the correct place or when API/ABI changes which with
> >>> such trashy code will happen unless development stops
> >> OK so the quickest path looks like to duplicate the code as it was
> >> done in the original version from Hemanth...
> > yes, i dont like it but i think this is a rare case where this will cause
> > fewer problems
> > also the code in libavcodec should be marked as deprecated as soon as possible
> > [...]
> To clarify:
> * The filter will wrap the existing av_audio_convert_alloc and
> av_audio_convert functions to do the sample format conversion.
> * It will duplicate (and extend) the code for channel mixing.
> Also, I was wondering if it makes sense to split this filter into two:
Michael GnuPG fingerprint: 9FF2128B147EF6730BADF133611EC787040B0FAB
Asymptotically faster algorithms should always be preferred if you have
asymptotical amounts of data
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Size: 198 bytes
Desc: Digital signature
More information about the ffmpeg-devel