[MPlayer-dev-eng] altivec patch 5/5: don't output the wrong format

[Alan Curry]

Diego Biurrun writes the following:
>
>On Wed, Feb 08, 2006 at 12:36:32PM +0100, Michael Niedermayer wrote:
>> 
>> patch ok, can be applied (by someone with PPC) though i somehow would
>> prefer if the switch() wherent in the innermost loop, but then its better
>> then having non-working code ...
>
>Applied.
>
>Alan, it would be nice if you could address Michael's comment in another
>patch.

OK, I can split altivec_yuv2packedX into separate functions, with a macro to
declare them all with nearly-identical bodies, following the precedent set by
DEFCSP420_CVT. The callers will then select the output format by calling the
specific relevant function, which also gives us a chance to handle the
formats that aren't supported in the altivec code by calling yuv2packedXinC
instead.

That way ends with 2 families of functions: the current DEFCSP420_CVT and the
new family. The main difference between them is that the new family does
resizing and the old family doesn't. (They also might not support the exact
same set of input and output formats...)