Troubleshooting common installation issues
Having trouble with mamba/conda or your Python installation? You’ve come to the right place. Below, we outline some of the most common issues encountered during local installation of Python packages, as well as the troubleshooting steps to try to fix the issues.
Common Problems
Below we list some of the most common issues encountered when installing packages with mamba.
Missing package errors
If you have successfully followed the steps outlined in the installation guide to create your environment, but receive an ImportError when importing packages, you might have launched the notebook from the 'base' environment instead of the ‘musa-550-fall-2023’ environment. Be sure to activate the ‘musa-550-fall-2023’ environment before launching the notebook.s
The file extension of the environment file on Windows
Be sure that Windows does not automatically add an .txt extension to your environment.yml file. This will sometimes cause mamba to fail, with a cryptic error:
SpecNotFound: environment with requirements.txt needs a nameThe environment file needs to end in .yml. You can change the extension for a file on Windows following these instructions.
Mixing pip and mamba
The command pip can also be used to install Python packages. However, using pip to install packages into a mamba environment can lead to issues. It’s best to stick to using the mamba env update command to update your environment or mamba install package_name to install specific packages.
Import errors for geopandas
When importing geopandas, you can sometimes receive errors about missing libraries. This is usually because package versions got mixed up during installation. This can sometimes happen, and geopandas is particularly sensitive to the versions of its dependencies.
The best and easiest thing to do to try to solve it is use the steps above to create a fresh environment.
Numpy errors
If you receive the following error:
ImportError: Something is wrong with the numpy installation. While importing we detected an older version of numpy in ['/path/to/old/version/of/numpy/'']. One method of fixing this is to repeatedly uninstall numpy until none is found, then reinstall this version.From the Miniforge Prompt (Windows) or Terminal (Mac), run:
mamba install --force-reinstall --clobber numpyMost common fix: install a fresh environment
Unfortunately, mamba/conda environments can sometimes become corrupted, preventing new packages from being installed or imported into Python properly. Most issues like this can be solved by simply deleting your current environment and starting fresh with a new version.
The following steps can be used to try to solve common issues:
The commands here should be executed via the command line, either using the Miniforge Prompt on Windows or the Terminal app on MacOS.
Step 1: Delete any existing environment
We want to create a fresh environment, so you can delete any environment that was giving you issues. If that environment was called ‘musa-550-fall-2023’, you can run the following commands to delete it:
mamba deactivatemamba env remove --name musa-550-fall-2023The first command will make sure the environment we are deleting isn’t active, and then the second command will perform the deletion.
Step 2: Create a fresh environment
Follow the instructions outlined here to create a fresh version of the course website.