You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
[56/359] Compiling C object numpy/_core/_struct_ufunc_tests.cpython-313-x86_64-linux-gnu.so.p/src_umath__struct_ufunc_tests.c.o
../numpy/_core/src/umath/_struct_ufunc_tests.c: In function ‘PyInit__struct_ufunc_tests’:
../numpy/_core/src/umath/_struct_ufunc_tests.c:160:5: warning: ‘Py_GIL_DISABLED’ is not defined, evaluates to ‘0’ [-Wundef]
160 | #if Py_GIL_DISABLED
| ^~~~~~~~~~~~~~~
many places. This is a choice by Python to not define the macro for builds without free threading, but we might want to put a guard on it just to avoid the noise.
../numpy/_core/include/numpy/npy_common.h:504:20: warning: ‘NPY_ULONG’ is not defined, evaluates to ‘0’ [-Wundef]
504 | #define NPY_UINT64 NPY_ULONG
| ^~~~~~~~~
../numpy/_core/src/multiarray/descriptor.c:2098:22: note: in expansion of macro ‘NPY_UINT64’
2098 | #if NPY_ULONGLONG == NPY_UINT64
| ^~~~~~~~~~
Probably came about when we dropped the old header. Should be defined in numpyconfig.h
[352/359] Compiling C object numpy/random/_common.cpython-313-x86_64-linux-gnu.so.p/meson-generated_numpy_random__common.pyx.c.o
numpy/random/_common.cpython-313-x86_64-linux-gnu.so.p/numpy/random/_common.pyx.c:2814:35: warning: ‘CYTHON_PROFILE’ is not defined, evaluates to ‘0’ [-Wundef]
2814 | #if CYTHON_USE_SYS_MONITORING && (CYTHON_PROFILE || CYTHON_TRACE)
Probably a Cython choice not to define the macro.
Also in a test:
limited_api_latest.c:1:23: error: ‘PY_VERSION_HEX’ is not defined, evaluates to ‘0’ [-Werror=undef]
1 | #if Py_LIMITED_API != PY_VERSION_HEX & 0xffff0000
| ^~~~~~~~~~~~~~
cc1: all warnings being treated as errors
Probably failure to import Python.h
The text was updated successfully, but these errors were encountered:
seberg
added a commit
to seberg/numpy
that referenced
this issue
Jun 17, 2025
As noted by Chuck in numpygh-29138, there are some undef warnings that seem not nice
this should fix them.
The fact that `NPY_LONG`, etc. are not defined at macro expansion time is a bit
of a trap, maybe it would be nice to have CI fail for this...
As noted by Chuck in gh-29138, there are some undef warnings that seem not nice
this should fix them.
The fact that `NPY_LONG`, etc. are not defined at macro expansion time is a bit
of a trap, maybe it would be nice to have CI fail for this...
Uh oh!
There was an error while loading. Please reload this page.
We are using undefined macros in several places.
many places. This is a choice by Python to not define the macro for builds without free threading, but we might want to put a guard on it just to avoid the noise.
Probably came about when we dropped the old header.Should be defined innumpyconfig.h
Probably a Cython choice not to define the macro.
Also in a test:
Probably failure to import Python.h
The text was updated successfully, but these errors were encountered: