Python: isolation and management
There are many ways to run and maintain your own python and/or python modules. Here are some tools and examples to do so.
Python
This level just uses the default python, or optionally installed python, that comes with the OS. You simply write a script and run it.
Examples
#!/usr/bin/python
or
#!/usr/bin/env python3.12
pip
Here, you need a python module that doesn't come with the version(s) of python installed with the OS. This will install the python module in your home account (E.g. ~/.local/lib/python3.9/site-packages).
pip install --user casatools
Python Virtual Environment (venv)
Creating a Python Virtual Environment allows other users to run your code without having to install their own copy of the needed python modules. The venv is just a directory. Below is an example that creates a venv called my-env
This will create, in your current directory, a directory named "example" that contains the virtual environment.
python -m venv example
To use a virtual environment, you must activate it by sourcing the activate script inside the environment.
source example/bin/activate
Once you have activated the virtual environment you can install python packages into it as if you were the administrator of the machine. This differs from using pip to install python packages into your account (see above) in that the new package only exists in the virtual environment, and not in all instances of python that you invoke outside of the virtual environment.
While the virtual environment is active your prompt will change to indicate you are inside a virtual environment.
(example)$
You may want to upgrade pip before you use it to install modules
pip install --upgrade pip
Then you can install modules. Here is an example installing numpy
pip install numpy
When you are finished with your virtual environment, you can deactivate it with the deactivate command.
deactivate
To delete a virtual environment, just delete its directory. Name your virtual environments wisely, so that you can remember what they are days, weeks, or even months later.
miniforge
This is useful if you need a version of python that is not provided by the OS, or you just want to isolate yourself from the version provided by the OS. It allows you to keep using the same version of python regardless of what version is on the OS. So when your IT staff updates the OS, and a different version of python gets installed, your code should still work.
Using miniforge is very similar to using "conda" but without any licensing concerns.
Download and install
https://conda-forge.org/download/
bash Miniforge3-$(uname)-$(uname -m).sh
or for the CLI enthusiasts
wget -O Miniforge3.sh "https://github.com/conda-forge/miniforge/releases/latest/download/Miniforge3-$(uname)-$(uname -m).sh"
bash Miniforge3.sh -b -p "${HOME}/conda"
Run the environment
source "${HOME}/conda/etc/profile.d/conda.sh"
It will install a default version of python. In 2025 this seems to be python-3.12. You can install a different version (e.g. python-3.10) like so
Install a different version of python with something like
conda create --name py3.10 python=3.10
Activate the py3.10 environment
conda activate py3.10
To deactivate the two environments
conda deactivate
conda deactivate
container
A container is useful when you need to make your code portable so that others not at the NRAO can use it.
