Python: pyenv, pyvenv, virtualenv – What’s the difference?

So I see questions around these terms very often in our growing Python Bangladesh community. Most of the times beginners are confused about what is what. I hope I can refer to this blog post to explain the similarities and differences.

pyenv

Have you ever wanted to test your code against multiple versions of Python? Or just wanted to install a newer version of Python without affecting your existing version? May be you heard about PyPy a lot and want to install it on your machine?

If you did, then pyenv is the perfect tool for you. It allows you to easily install multiple copies and multiple flavors of the Python interpreter. So you can not only install different versions of CPython, you can also install PyPy, Jython, Stackless Python and their different versions.

The tool provides a nice command line tool to easily swap out the global python interpreter. It also allows to define per application python version. You can use it’s local command or directly mention a python version in a file named .python-version under a directory and for that directory and it’s children, the mentioned version will be used.

Trust me, this project is awesome. I use it to switch between Python 2 and 3 on my local machine. I also use it often on servers to quickly install any flavor/version of Python. Do check out their docs, you will love it.

pyvenv & virtualenv

pyvenv and virtualenv allow you to create virtual environments so we can isolate our project dependencies. Why are they helpful? Say for example, you have one project which uses Django 1.6 still while your newer projects start with 1.9. When you install one version of Django, it replaces the other one, right? Virtual environments can rescue us from such situation. From the official docs:

A virtual environment (also called a venv) is a Python environment such that the Python interpreter, libraries and scripts installed into it are isolated from those installed in other virtual environments, and (by default) any libraries installed in a “system” Python, i.e. one which is installed as part of your operating system.

When we create a new virtual environment, it creates an isolated environment with it’s own local interepreter linked to it’s own libraries/scripts paths. So when we use this local interpreter, it loads the libraries from the local environment. If it can’t find one locally, then tries to locate that library in the parent/system environment.

Please note, these tools do not compile/install new Python interpreters. They simply create “virtual environments” on top of an installed Python version. Say, I have Python 3.5 installed on my machine and created virtual environments for this version. Then these environments would also have local copies of Python 3.5, except their environment paths would point to different locations. It’s like we’re copying the main interpreter to a new location and then making it use a different path to load libraries and packages.

virtualenv is often the most popular choice for creating the virtual environments. It has been around for a long period of time, it supports Python versions from 2.6 up to the latest 3.5. But it’s not something built into the standard Python distribution. You have to install it from the PyPi.

pyvenv comes with Python standard distribution from version 3.4. There is also a venv module in the standard library which allows us to access this functionality programmatically. We can find more details here: https://docs.python.org/3/library/venv.html.

Summary

pyenv – A Python version manager. Installs different versions and flavors of Python interpreters.

pyvenv – A tool to create isolated virtual environments from a Python interpreter. Ships with Python from 3.4.

virtualenv – Creates virtual environments, available in PyPi.

So pyvenv is comparable to virtualenv while pyenv is a totally different kind of tool.