Hi everyone, I tried to build numpy 0.9.8 on compaq alpha tru64 UNIX v5.1 with gcc 4.0.2, but I encounterd the compilation trouble. The error message is the following. Do you have any suggestion? Thank you in advance. Shimizu. numpy/core/src/umathmodule.c.src: In function 'nc_floor_quotl': numpy/core/src/umathmodule.c.src:600: warning: implicit declaration of function 'floorl' numpy/core/src/umathmodule.c.src:600: warning: incompatible implicit declaration of built-in fu nction 'floorl' .... numpy/core/src/umathmodule.c.src: In function 'LONGDOUBLE_floor_divide': numpy/core/src/umathmodule.c.src:1050: warning: incompatible implicit declaration of built-in f unction 'floorl' numpy/core/src/umathmodule.c.src: In function 'CLONGDOUBLE_absolute': numpy/core/src/umathmodule.c.src:1319: warning: incompatible implicit declaration of built-in f unction 'sqrtl' .... build/src.osf1-V5.1-alpha-2.4/numpy/core/__umath_generated.c: At top level: build/src.osf1-V5.1-alpha-2.4/numpy/core/__umath_generated.c:15: error: 'acosl' undeclared here (not in a function) build/src.osf1-V5.1-alpha-2.4/numpy/core/__umath_generated.c:18: error: 'acoshf' undeclared her e (not in a function) ...
N Shimizu wrote:
Hi everyone,
I tried to build numpy 0.9.8 on compaq alpha tru64 UNIX v5.1 with gcc 4.0.2,
but I encounterd the compilation trouble.
Thanks for the test. This looks like a configuration problem. Could you post the config.h file that is generated when you run python setup.py It should be found in build/src.<platform>-<version>/numpy/core/config.h I don't think we've got the right set of configurations going for that platform. Basically, we need to know if it has certain float and long versions of standard math functions (like floorf and floorl). It looks like the configuration code detected that it didn't have these functions but then during compilation, the functions that NumPy created were already defined causing the error. If we can first get a valid config.h file for your platform, then we can figure out how to generate it during build time. -Travis
Thank you for your reply.
The "config.h" is the following.
I hope it will be helpful.
Shimizu
/* #define SIZEOF_SHORT 2 */
/* #define SIZEOF_INT 4 */
/* #define SIZEOF_LONG 8 */
/* #define SIZEOF_FLOAT 4 */
/* #define SIZEOF_DOUBLE 8 */
#define SIZEOF_LONG_DOUBLE 16
#define SIZEOF_PY_INTPTR_T 8
/* #define SIZEOF_LONG_LONG 8 */
#define SIZEOF_PY_LONG_LONG 8
/* #define CHAR_BIT 8 */
#define MATHLIB m
#define HAVE_LONGDOUBLE_FUNCS
#define HAVE_FLOAT_FUNCS
#define HAVE_LOG1P
#define HAVE_EXPM1
#define HAVE_INVERSE_HYPERBOLIC
#define HAVE_INVERSE_HYPERBOLIC_FLOAT
#define HAVE_INVERSE_HYPERBOLIC_LONGDOUBLE
#define HAVE_ISNAN
#define HAVE_RINT
2006/6/30, Travis Oliphant
N Shimizu wrote:
Hi everyone,
I tried to build numpy 0.9.8 on compaq alpha tru64 UNIX v5.1 with gcc 4.0.2,
but I encounterd the compilation trouble.
Thanks for the test. This looks like a configuration problem. Could you post the config.h file that is generated when you run python setup.py
It should be found in
build/src.<platform>-<version>/numpy/core/config.h
I don't think we've got the right set of configurations going for that platform. Basically, we need to know if it has certain float and long versions of standard math functions (like floorf and floorl).
It looks like the configuration code detected that it didn't have these functions but then during compilation, the functions that NumPy created were already defined causing the error.
If we can first get a valid config.h file for your platform, then we can figure out how to generate it during build time.
-Travis
participants (2)
-
N Shimizu -
Travis Oliphant