[BUG]: Error importing numpy: you should not try to import numpy from its source directory;

[danny-su]

Bug description

Steps to reproduce

$ mojo
Welcome to Mojo! 🔥

Expressions are delimited by a blank line.
Type `:quit` to exit the REPL and `:mojo help` for further assistance.

  1> from python import Python
  2. let np = Python.import_module("numpy")
  3. let array = np.array([1, 2, 3])
  4. print(array)
  5.
Error: Error importing numpy: you should not try to import numpy from
        its source directory; please exit the numpy source tree, and relaunch
        your python interpreter from there.
  5>

System information

- What OS did you do install Mojo on ?
macOS 14.3.1
- Provide version information for Mojo by pasting the output of `mojo -v`
mojo 0.7.1 (af002202)
- Provide Modular CLI version by pasting the output of `modular -v`
modular 0.5.0 (47924ae4)

[catalpaaa]

Same issue here on x86_64 ubuntu 23.10 :(

python_lib = /home/catalpa/miniforge3/envs/a3/lib/libpython3.12.so

[danny-su]

@catalpaaa Removing Mojo completely and then reinstalling it will fix this issue.

[kseuro]

I have the same issue on a MacOS Ventura 13.4. This error persists even after completely removing mojo and modular, and reinstalling everything.

Output of mojo -v mojo 24.1.0 (55ec12d6)

Output of modular -v modular 0.5.1 (1b608e3d)

I created a conda environment using

conda create -yn mojo python=3.11

In my bash_profile, the following paths are set:

export MODULAR_HOME="$HOME/.modular"
export PATH="$MODULAR_HOME/pkg/packages.modular.com_mojo/bin:$PATH"
export MOJO_PYTHON_LIBRARY="$HOME/miniconda3/envs/mojo/lib/libpython3.11.dylib"

The easiest way for me to reproduce this error is to run the RayTracing.ipynb under the examples directory in this repo. Calling the render function throws:

Error: Error importing numpy: you should not try to import numpy from
        its source directory; please exit the numpy source tree, and relaunch
        your python interpreter from there.

[kseuro]

I have the same issue on a MacOS Ventura 13.4. This error persists even after completely removing mojo and modular, and reinstalling everything.

Output of mojo -v mojo 24.1.0 (55ec12d6)

Output of modular -v modular 0.5.1 (1b608e3d)

I created a conda environment using

conda create -yn mojo python=3.11

In my bash_profile, the following paths are set:

export MODULAR_HOME="$HOME/.modular"
export PATH="$MODULAR_HOME/pkg/packages.modular.com_mojo/bin:$PATH"
export MOJO_PYTHON_LIBRARY="$HOME/miniconda3/envs/mojo/lib/libpython3.11.dylib"

The easiest way for me to reproduce this error is to run the RayTracing.ipynb under the examples directory in this repo. Calling the render function throws:

Error: Error importing numpy: you should not try to import numpy from
        its source directory; please exit the numpy source tree, and relaunch
        your python interpreter from there.

I figured out that I am able to execute the ray tracing code as a stand-alone mojo script using iterm but NOT vscode's built-in terminal. This is will be my work around for the time being.

[andompesta]

same issue for me, after updateing mojo I can't import numpy anymore. I can't event import from iTerm.

numpy version = 1.23 mojo version = 24.1.0 modular version = 0.5.2 conda version = 23.11.0 macOs = 14.2.1

system path:

'/opt/homebrew/opt/[email protected]/Frameworks/Python.framework/Versions/3.12/lib/python3.12/site-packages', 
'~/opt/miniconda3/envs/env_name/lib/python310.zip',
 '~/opt/miniconda3/envs/env_name/lib/python3.10',
 '~/opt/miniconda3/envs/env_name/lib/python3.10/lib-dynload', 
'~/opt/miniconda3/envs/env_name/lib/python3.10/site-packages'

[rafelafrance]

For me it was caused by a mismatch between the system python v3.10 and my virtual environment python v3.11. If I set MOJO_PYTHON_LIBRARY it worked just fine. You'll need to hunt down your python libraries for your system. For me it was:

export MOJO_PYTHON_LIBRARY=/usr/lib/x86_64-linux-gnu/libpython3.11.so.1.0 

[andompesta]

For me it was caused by a mismatch between the system python v3.10 and my virtual environment python v3.11. If I set MOJO_PYTHON_LIBRARY it worked just fine. You'll need to hunt down your python libraries for your system.

Indeed I manage to fix it after updating my conda environment to python 3.12 as it matches the mojo python_lib version.