[this time as plain text oops] > Yes, it's been considered. It's just hard, and few people would > benefit. Fair enough. At least I'm not crazy. :) (more coming on the ELF/visibility thread, but I have to do some experiments) > gmp/mpfr/mpc Right, I spent some hours banging my head against this. It "doesn't work" -- using gcc 4.0.1 on MacOSX. Whereas letting gcc build do it does. I eventually gave up...and figured it out, at about the same time. Within gcc, the -std=gnu99 is omitted, because CFLAGS=-g -O2 gets passed down, avoiding problems in Apple's gcc 4.0.1. There is also a small patch to gmp-h.in that can do it. Seems like many people have run into this and generally they just use CC=gcc-4.2 to get around it. like gmp-h.in: #ifdef __GNUC__ > /* Apple gcc 4.0.1 -std=gnu99 has problems, worked around here. */ > #if defined __GNUC_STDC_INLINE__ && defined __APPLE__ && __GNUC__ == 4 && __GNUC_MINOR__ == 0 > #define __GMP_EXTERN_INLINE __inline__ > #define __GMP_INLINE_PROTOTYPES 0 / * This is the part that I didn't think to try for a while. */ > #else #if (defined __GNUC_STDC_INLINE__) || (__GNUC__ == 4 && __GNUC_MINOR__ == 2) #define __GMP_EXTERN_INLINE extern __inline__ __attribute__ ((__gnu_inline__)) #else #define __GMP_EXTERN_INLINE extern __inline__ #endif #define __GMP_INLINE_PROTOTYPES 1 #endif > #endif I have mail with the gmp folks. - Jay [snip]
