Neil Schemenauer wrote: > In gmane.comp.python.python-3000.cvs, you wrote: >> Added: python/branches/py3k/Objects/doubledigits.c >> ============================================================================== >> --- (empty file) >> +++ python/branches/py3k/Objects/doubledigits.c Mon Dec 10 23:19:17 2007 >> @@ -0,0 +1,601 @@ >> +/* Free-format floating point printer >> + * >> + * Based on "Floating-Point Printer Sample Code", by Robert G. Burger, >> + * http://www.cs.indiana.edu/~burger/fp/index.html >> + */ >> + >> +#include "Python.h" >> + >> +#if defined(__alpha) || defined(__i386) || defined(_M_IX86) || >> defined(_M_X64) || defined(_M_IA64) >> +#define LITTLE_ENDIAN_IEEE_DOUBLE >> +#elif !(defined(__ppc__) || defined(sparc) || defined(__sgi) || >> defined(_IBMR2) || defined(hpux)) >> +#error unknown machine type >> +#endif > > gcc -pthread -c -fno-strict-aliasing -DNDEBUG -g -O3 -Wall > -Wstrict-prototypes -I. -IInclude -I../Include -DPy_BUILD_CORE -o > Objects/doubledigits.o ../Objects/doubledigits.c > ../Objects/doubledigits.c:12:2: error: #error unknown machine type > make: *** [Objects/doubledigits.o] Error 1 > > That's on a Debian Linux x86_64 machine. That code looks like a > portability nightmare. Maybe a autoconfig test would work better > (e.g. put a known number into a memory and inspect it).
Yeah, I hit the same spot when I tried to compile the code for the Win x64 platform. Do you have experience with autoconf and can you come up with a test? Christian _______________________________________________ Python-3000-checkins mailing list [email protected] http://mail.python.org/mailman/listinfo/python-3000-checkins
