new gfortran 4.1 branch failures

[Steve Kargl]

On Thu, Apr 20, 2006 at 06:16:37PM -0400, Jack Howarth wrote:
> Steve,
>    In libgfortran/generated, I have...
> 
> howarth% ls resh*
> reshape_c10.c   reshape_c4.c    reshape_i16.c   reshape_i8.c
> reshape_c16.c   reshape_c8.c    reshape_i4.c    reshape_r10.c
> howarth% ls trans*
> transpose_c10.c transpose_c4.c  transpose_i16.c transpose_i8.c
> transpose_c16.c transpose_c8.c  transpose_i4.c  transpose_r10.c
> 
> Also if I go into darwin_objdir/powerpc-apple-darwin8/libgfortran
> and execute...
> 
> howarth% ls resh* trans*
> reshape_c10.lo          reshape_i8.lo           transpose_c8.lo
> reshape_c10.o           reshape_i8.o            transpose_c8.o
> reshape_c16.lo          reshape_packed.lo       transpose_generic.lo
> reshape_c16.o           reshape_packed.o        transpose_generic.o
> reshape_c4.lo           reshape_r10.lo          transpose_i16.lo
> reshape_c4.o            reshape_r10.o           transpose_i16.o
> reshape_c8.lo           transfer.lo             transpose_i4.lo
> reshape_c8.o            transfer.o              transpose_i4.o
> reshape_generic.lo      transpose_c10.lo        transpose_i8.lo
> reshape_generic.o       transpose_c10.o         transpose_i8.o
> reshape_i16.lo          transpose_c16.lo        transpose_r10.lo
> reshape_i16.o           transpose_c16.o         transpose_r10.o
> reshape_i4.lo           transpose_c4.lo
> reshape_i4.o            transpose_c4.o
> 
> and try 'nm *.o | grep __gfortran_reshape_16' or 
> 'nm *.o | grep __gfortran_transpose_i16' I get no symbols found.
> Does that show any symbols for you?

I have

troutmask:sgk[210] nm *.o | grep gfortran_reshape
0000000000000000 T _gfortran_reshape_c10
0000000000000000 T _gfortran_reshape_c4
0000000000000000 T _gfortran_reshape_c8
00000000000006d0 T _gfortran_reshape
00000000000006b0 T _gfortran_reshape_char
0000000000000000 T _gfortran_reshape_16
0000000000000000 T _gfortran_reshape_4
0000000000000000 T _gfortran_reshape_8
0000000000000000 T _gfortran_reshape_r10
troutmask:sgk[211] nm *.o | grep gfortran_transpose
0000000000000000 T _gfortran_transpose_c10
0000000000000000 T _gfortran_transpose_c4
0000000000000000 T _gfortran_transpose_c8
0000000000000230 T _gfortran_transpose
0000000000000220 T _gfortran_transpose_char
0000000000000000 T _gfortran_transpose_i16
0000000000000000 T _gfortran_transpose_i4
0000000000000000 T _gfortran_transpose_i8
0000000000000000 T _gfortran_transpose_r10

Note, the single leading underscore.  IIRC the double leading
underscore is for builtin functions that should be expanded
by the backend.

Does darwin have both real(10) and real(16)?

-- 
Steve