Can't build 32-bit GMP on OS X 10.5.6

Wade Williams wadesworld at
Wed Jan 14 23:45:59 CET 2009

On Jan 14, 2009, at 16:26, Torbjorn Granlund wrote:

> Wade Williams <wadesworld at> writes:
>  I'm trying to build 32-bit GMP on OS X and getting the following  
> error:
>  Making all in mpn
>  /bin/sh ../libtool --mode=compile gcc -DHAVE_CONFIG_H -I. -I. -I.. -
>  D__GMP_WITHIN_GMP -I.. -DOPERATION_`echo fib_table | sed 's/_$//'`  -
>  DNO_ASM  -O3 -c -o fib_table.lo fib_table.c
>  mkdir .libs
>    gcc -DHAVE_CONFIG_H -I. -I. -I.. -D__GMP_WITHIN_GMP -I.. -
>  DOPERATION_fib_table -DNO_ASM -O3 -c fib_table.c  -fno-common -DPIC -
>  o .libs/fib_table.o
>  In file included from ../gmp-impl.h:102,
>                    from fib_table.c:4:
>  ../fib_table.h:4: error: syntax error before 'data'
>  fib_table.c:7: error: syntax error before 'data'
> You might want to check out line 4 in fib_table.h.
> I bet it says something like
>  Error, error, this data is for 64 bits
> meaning that you probably forgot to "make distclean" between
> incompatible builds.

Wonderful!  Thanks so much.    It compiles without a hitch on Intel  
and PPC now.

I saw the error in fib_table.h - but to look at that one line and  
automatically realize the answer is "make distclean"  rather than  
something else in the hundreds of source files, configure options or  
Apple compiler bugs.  I fully realize that to anyone with any  
reasonable amount of autotools experience, that's as plain as the nose  
on your face.

As a low-knowledge level user of the autotools, I have always  
defaulted to "make clean" thinking that "make distclean" was for  
package maintainers.  I see the error in that thinking now, and had I  
thought about it more, I probably would have said "Duh."

However, I may not be alone in such thinking.

A suggestion:

Perhaps additional dumb questions could be avoided with something like:

"Error, error, this data is for 64 bits - have you run make distclean  
between builds?"

Just a suggestion - thanks again!


More information about the gmp-discuss mailing list