[MPlayer-dev-eng] [PATCH] [TEST FUNC] Multi-channel reorder function

[Giacomo Comes]

On Sat, Nov 10, 2007 at 10:10:37PM +0800, Ulion wrote:
> Hello,
> 
> This is a reply to a post one year before:
> http://lists.mplayerhq.hu/pipermail/mplayer-dev-eng/2006-November/047404.html
> 
> We have this channel layout problem (different channel order between
> ad,ae,ao) for multi-channel audio for quite a long time. I'd like to
> push this forward and try to resolve the problem in some way.
> 
> Reimar Döffinger wrote:
> > On Tue, Nov 14, 2006 at 04:27:59PM -0800, Corey Hickey wrote:
> > > I don't know if it can be simplified. I see a couple things that could
> > > be different, but not necessarily better. In any case, the complexity of
> > > (a) is centralized, versus being distributed into many separate files
> > > (and only slightly lessened in most cases). Adding support to an
> > > ad/ae/ao is much easier with my approach.
> >
> > well, the complexity maybe centralized with the af approach, but it is
> > also far away from the real problem, probably making it even harder to
> > make a real fix.
> > I don't know. I think I mostly dislike inserting a filter, and that the
> > fix seems so far away from the problem. I wonder how complicated it
> > would be to instead create a reorder function that the ao/ad/aes can
> > use. Or at least in some other was encapsulate that new code in a
> > separate function instead of being "hidden" deep in the bowels of filter
> > chain creation.
> >
> > Greetings,
> > Reimar Döffinger
> >
> 
> I wrote such a reorder function that the ao/ad/aes can use it to
> reorder channels with an optimized speed (the copy func is about
> 1.5-1.6 times of memcpy on my PPC). By now the function can used to
> replace memcpy for 5/6 channel audios. Also another reorder function
> work on one buffer is also here.
> 
> This patch is just to show the idea, could be buggy and ugly, maybe
> disliked by somebodies, it's ok, I just want to fix the aac wrong
> channel order bug, wellcome any idea to resolve the problem. Shall we
> end this bug in the near future by either way?

I agree to end the bug. We have now three different patches that address
the same problem. I hope that finally the developers can agree on one of
them so we can check/test/apply the selected one.
I have worked on one of the patches but I don't care which one is choosen,
the important is that finally we can close the bug.

Ciao
Giacomo