All the python environment setup you need (probably)

2020-10-05

Edit (2022): I spent some time with Poetry, and would now recommend that for larger or long-lived projects.

Python environments are tricky, but they don’t need to be so tricky. Here is all the setup you need for many projects.

python3 -m venv .venv
source .venv/bin/activate

This will create a virtual environment called .venv (I say “vee-env”) in the directory in which it is run. It is a good idea to localize environments to projects, but you should not commit the venv folder to git (it will grow large).

Unfortunately, even with a venv, python dependency management is still a bit cumbersome. You can keep dependencies in a requirements.txt file, but doing so requires that you curate that file. If you need something to be reproducible quickly, simply dumping it with

pip freeze > requirements.txt

works, but will give you an unnecessarily large list. Much better is to manually list what you install by adding it with version number to a requirements file, so that your file looks something like this:

pandas==1.1.2
numpy==1.19.2
matplotlib==3.3.2

If you do use venvs as suggested above, I strongly recommend including something like this little snippet of Ines Montani’s in your .bashrc/ .zshrc / .shell-of-your-choice-rc:

function venv {
    if [ ! -d ".venv" ]; then
        python3 -m venv .venv
        echo -e "[38;5;2m✔ Created virtualenv .env[0m"
    fi
    source .venv/bin/activate
    export PYTHONPATH=`pwd`
    echo -e "[38;5;2m✔ Activated virtualenv .env[0m"
}

alias venv="venv"

With that snippet, you can create (if none exists) or activate (if one already exists) a venv in whatever directory you are in with just the venv command.

Thanks for reading. If you'd like words like these in your inbox, sign up below. Or just check this site occasionally. Or run, now, and never speak of this place. You do you.