A Python virtualenv is a Python interpreter and set of installed Python packages. Python packages are installed separate from the main system so you don’t need to use
su or worry about installing things system-wide. Since the interpreter (the
python run) and packages in a virtualenv are separate from other virtualenvs, you can switch between different versions of Python and different versions of installed packages with a single command.
Using virtualenv lets you do things like:
- Replicate your production Python environment in a dev setup so you can be sure you’re writing code and tests using the same package versions your deployed code will use.
- Create environments with the same set of Python packages but using different versions of the Python interpreter (e.g., Python 2.5, Python 2.7, and PyPy).
- Setup experimental environments for trying out new Python package versions or new Python software projects.
The easiest way to get going with virtualenv on Mac and Linux is to use the virtualenv-burrito installer. This one-line command installs virtualenv and virtualenvwrapper (a nice way to use virtualenvs):
curl -s https://raw.github.com/brainsik/virtualenv-burrito/master/virtualenv-burrito.sh | $SHELL
Once installed, you can make new virtualenvs with
mkvirtualenv <name>, install packages with
pip install <package>, and switch between virtualenvs using
Let’s make a virtualenv and run the example Sauce Labs test:
mkvirtualenv saucelabs pip install selenium
curl -s https://saucelabs.com/example/se2/python/private-?????? | bash
To install the same Python package versions in another virtualenv we can use
pip freeze to get what’s in our current environment, save it to a file, and use that file as an install list:
workon saucelabs pip freeze > requirements.txt mkvirtualenv likesaucelabs pip install -r requirements.txt
By developing, testing, and running production Python code in virtualenvs created using the same requirements files, you greatly reduce the risk of writing bugs which only show up in one environment but not another. We highly recommend using this tool to help ship code faster.