install location not inside site-packages
I noticed that h3-py was not installed correctly into the site-packages/, which causes a bunch of h3-related source files in a "root" python folder. This happened when I tried to remove packages (pip uninstall h3):
Found existing installation: h3 3.7.7
Uninstalling h3-3.7.7:
Would remove:
/home/wklee/.pyenv/versions/3.12.2/envs/py312/CHANGELOG.md
/home/wklee/.pyenv/versions/3.12.2/envs/py312/CMakeLists.txt
/home/wklee/.pyenv/versions/3.12.2/envs/py312/LICENSE
/home/wklee/.pyenv/versions/3.12.2/envs/py312/include/h3/h3api.h
/home/wklee/.pyenv/versions/3.12.2/envs/py312/lib/cmake/h3/h3Config.cmake
...
However, other packages (e.g., pip uninstall numpy) give
Found existing installation: numpy 1.26.4
Uninstalling numpy-1.26.4:
Would remove:
/home/wklee/.pyenv/versions/3.12.2/envs/py312/bin/f2py
/home/wklee/.pyenv/versions/3.12.2/envs/py312/lib/python3.12/site-packages/numpy-1.26.4.dist-info/*
/home/wklee/.pyenv/versions/3.12.2/envs/py312/lib/python3.12/site-packages/numpy.libs/libgfortran-040039e1.so.5.0.0
/home/wklee/.pyenv/versions/3.12.2/envs/py312/lib/python3.12/site-packages/numpy.libs/libopenblas64_p-r0-0cf96a72.3.23.dev.so
/home/wklee/.pyenv/versions/3.12.2/envs/py312/lib/python3.12/site-packages/numpy.libs/libquadmath-96973f99.so.0.0.0
/home/wklee/.pyenv/versions/3.12.2/envs/py312/lib/python3.12/site-packages/numpy/*
I tried this on different environments, including my windows env and still see it..
Hmmm, good catch @wingkitlee0. That's weird. Off the top of my head, I'm not sure why that's happening, but it does seem wrong.
I can try to take a look when I have time, but I'd be more than happy if anyone else looks into this in the meantime. :)
This is an old bug in skbuild related to MANIFEST.in.
https://github.com/scikit-build/scikit-build/issues/680
which seems to be not fixed fully (PR was reverted)
In other packages, they simply use package_data directly: https://github.com/Qiskit/qiskit-aer/pull/1677
P.S. I noticed this when I was trying to grab the h3api.h ..
Nice. Would you be willing to put up a PR to fix?
In the long term, I think we could also try to modernize the build, as in https://github.com/uber/h3-py/issues/340
Sure, I can try.
I believe this is resolved.