[Python] why do i have this compile error installing numpy in venv on cygwin?

I am running python 3.9 in cygwin on Windows 10. The python is cygwin's python. I tried "pip install numpy" from a newly created venv. However it was not able to find a compatible wheel and instead started to build numpy. Unfortunately, it died compiling several .c files on account of sys/select.h missing. FWIW, chatgpt thinks this is because the build system (Ninja?) thinks that its still trying to build for cygwin instead of natively (even though I am using ming32-gcc). Below is a sample of one of the errors

...

[194/535] Generating numpy/_core/_multiarray_tests.cpython-39-x86_64-cygwin.dll.p/_multiarray_tests.c

[195/535] Compiling C object numpy/_core/libnpymath.a.p/meson-generated_ieee754.c.o
FAILED: numpy/_core/libnpymath.a.p/meson-generated_ieee754.c.o
/usr/bin/x86_64-w64-mingw32-gcc -Inumpy/_core/libnpymath.a.p -Inumpy/_core -I../numpy/_core -Inumpy/_core/include -I../numpy/_core/include -I../numpy/_core/src/npymath -I../numpy/_core/src/common -I/usr/include/python3.9 -I/tmp/pip-install-2nk4smdu/numpy_d0c3008da8224ebc9f1bede0e4cba273/.mesonpy-y0j8jkqq/meson_cpu -fdiagnostics-color=always -DNDEBUG -D_FILE_OFFSET_BITS=64 -Wall -Winvalid-pch -std=c11 -O3 -fno-strict-aliasing -msse -msse2 -msse3 -DNPY_HAVE_SSE2 -DNPY_HAVE_SSE -DNPY_HAVE_SSE3 -MD -MQ numpy/_core/libnpymath.a.p/meson-generated_ieee754.c.o -MF numpy/_core/libnpymath.a.p/meson-generated_ieee754.c.o.d -o numpy/_core/libnpymath.a.p/meson-generated_ieee754.c.o -c numpy/_core/libnpymath.a.p/ieee754.c
In file included from /usr/include/python3.9/Python.h:50,
from ../numpy/_core/src/npymath/npy_math_common.h:4,
from ../numpy/_core/src/npymath/ieee754.c.src:7:
/usr/include/python3.9/pyport.h:230:10: fatal error: sys/select.h: No such file or directory

230 | #include <sys/select.h>
| ^~~~~~~~~~~~~~
compilation terminated.


Any ideas what the problem is?

What's bizarre is that at the top level (outside the venv) I have numpy already installed, although I cannot recall how it got there - cygwin setup shows that I didn't install it from cygwin, and a recent attempt to upgrade numpy lead to the same error as above.

Continue reading...