Python virtual environments - what you need to know

When we’re working on a project, we will most likely need to leverage Python packages found on PyPi or conda forge.

While you can install these packages directly onto your machine, it may only work for that one specific project. Over time, packages update and become incompatible with dependencies from a previous versions.

What is a virtual environment?

A virtual environment is a tool where you can isolate project dependencies. You can install different versions of packages (and even Python itself) without it affecting other projects.

In the above example, our 3 projects have different dependencies. Project 1 and 2 share flask as a common dependency, but project 3 doesn’t. Moreover, Project 1 doesn’t require matplotlib, but Project 3 does.

Virtual environments allow us to define these dependencies independently of other.

Creating a virtual environment

There are a few ways to create a virtual environment. The first is by using venv, a Python built-in that allows you to create these with one command. Another is by using Conda.

For the purposes of this, we’ll be sticking with venv, as conda can get quite bulky (unless you use miniconda or mamba). To create a virtual environment using venv, navigate to the root of your project and type the following command:

python venv -m <environment name>

Where <environment name> is whatever you’d like to call the environment.

Activating your virtual environment

Now that the virtual environment is created, we need to be able to use it. It’s operating system dependent:

• Windows: <environment name>\Scripts\activate.bat

• Mac/Linux: source <environment name>/bin/activate

Where <environment name> is the name of your virtual environment.

Installing packages into a virtual environment

To install packages, you’ll be leveraging pip. Python’s venv module comes bundled together with pip. You can install any package listed in PyPi using pip. To do so:

pip install <package>

Where <package> is the name of the package you want to install. You can get a bit crafty with specific versions:

pip install <package>==<version>

Where <version> is the version you want to install. For instance, if we wanted to install flask 2.0.1:

pip install flask==2.0.1

» Fun tidbit: pip stands for “pip install python”