Support the MP library on recent FreeBSD, have it pass relevant tests. src/ChangeLog addition: 2011-11-26 Aidan Kehoe <kehoea@parhasard.net> * number-mp.c (bignum_to_string): Don't overwrite the accumulator we've just set up for this function. * number-mp.c (BIGNUM_TO_TYPE): mp_itom() doesn't necessarily do what this code used to think with negative numbers, it can treat them as unsigned ints. Subtract numbers from bignum_zero instead of multiplying them by -1 to convert them to their negative equivalents. * number-mp.c (bignum_to_int): * number-mp.c (bignum_to_uint): * number-mp.c (bignum_to_long): * number-mp.c (bignum_to_ulong): * number-mp.c (bignum_to_double): Use the changed BIGNUM_TO_TYPE() in these functions. * number-mp.c (bignum_ceil): * number-mp.c (bignum_floor): In these functions, be more careful about rounding to positive and negative infinity, respectively. Don't use the sign of QUOTIENT when working out out whether to add or subtract one, rather use the sign QUOTIENT would have if arbitrary-precision division were done. * number-mp.h: * number-mp.h (MP_GCD): Wrap #include <mp.h> in BEGIN_C_DECLS/END_C_DECLS. * number.c (Fbigfloat_get_precision): * number.c (Fbigfloat_set_precision): Don't attempt to call XBIGFLOAT_GET_PREC if this build doesn't support big floats.