python-scikit-learn can't be imported "module compiled against API version 0x10 but this version of numpy is 0xf"

Dear Guix maintainers,

If I run:

guix shell python-scikit-learn

and in it, run:

python3 -c "from sklearn.feature_extraction.text import CountVectorizer"

I get:

RuntimeError: module compiled against API version 0x10 but this version of numpy is 0xf . Check the section C-API incompatibility at the Troubleshooting ImportError section at https://numpy.org/devdocs/user/troubleshooting-importerror.html#c-api-incompatibilityfor indications on how to solve this problem .

Traceback (most recent call last):

File "<string>", line 1, in <module>

File "/usr/lib/python3.10/site-packages/sklearn/__init__.py", line 82, in <module>

from .base import clone

File "/usr/lib/python3.10/site-packages/sklearn/base.py", line 17, in <module>

from .utils import _IS_32BIT

File "/usr/lib/python3.10/site-packages/sklearn/utils/__init__.py", line 17, in <module>

from scipy.sparse import issparse

File "/usr/lib/python3.10/site-packages/scipy/sparse/__init__.py", line 267, in <module>

from ._csr import *

File "/usr/lib/python3.10/site-packages/scipy/sparse/_csr.py", line 10, in <module>

from ._sparsetools import (csr_tocsc, csr_tobsr, csr_count_blocks,

ImportError: numpy.core.multiarray failed to import

I can't look into it right now, but I'm available to provide any

necessary information.

Cheers,

Edouard.

I found the cause:

.local/lib/python3.10/site-packages/ is part of the interpreter's load

path, and so an old version of numpy was imported in priority over

guix's.

I don't know if this is by design, but if so, it means that any

pip --user installed package will overshadow any guix install package,

even when giving --pure to guix shell. It may not be what we want.

What made me find this is that sklearn worked when in a container, but

not in a shell, that meant that something from the local system was

leaking into the shell.

help-debbugs@gnu.org (GNU bug Tracking System) writes: