[MPlayer-dev-eng] Problem compiling mplayer for PowerPC + AltiVec

[Paul TT]

On Thu, 7 Jul 2005 01:05:02 +0200
Diego Biurrun <diego at biurrun.de> wrote:

[...]
> > ppc/idct_altivec.c:160: error: initializer element is not constant
> > ppc/idct_altivec.c:160: error: (near initialization for 'constants[0]')
> > ppc/idct_altivec.c:161: error: initializer element is not constant
> > ppc/idct_altivec.c:161: error: (near initialization for 'constants[1]')
> > ppc/idct_altivec.c:162: error: initializer element is not constant
> > ppc/idct_altivec.c:162: error: (near initialization for 'constants[2]')
> > ppc/idct_altivec.c:163: error: initializer element is not constant
> > ppc/idct_altivec.c:163: error: (near initialization for 'constants[3]')
> > ppc/idct_altivec.c:165: error: initializer element is not constant
> > ppc/idct_altivec.c:165: error: (near initialization for 'constants[4]')
> > make[1]: *** [ppc/idct_altivec.o] Error 1
> > make[1]: Leaving directory `/tmpfs/mplayer/libavcodec'
> > make: *** [libavcodec/libavcodec.a] Error 2
> > ========

[...]
> 
> Yes, I've noticed this a while ago with gcc 3.3.5 on Debian Sarge.  The
> strange thing is that FFmpeg (without ffplay) compiles fine.
> 

same problem here :-((((
the problem it's in the AAV macro located in libavcodec/ppc/gcc_fixes.h, which seems not to work with gcc and altivec
i posted the problem a few weeks ago, then i substituted every AAV(x,x,x,x) (by hand) specs in source files which use it, then the compilation was successfully....

i didnt' post a solution since mine it's not a solution ;-)

the real question is:

WHY is there this AAV declaration??????? it's not better off to declare directly all those arrays as constant arrays?!?!?!?!?!?!

if this should be ok, i'll post a patch doing this....