Getting Started with virtualenv

Virtual environments can be really handy for testing software. That's true in programming circles too. Ian Bicking created the virtualenv project, which is a tool for creating isolated Python environments. You can use these environments to test out new versions of your software, new versions of packages you depend on or just as a sandbox for trying out some new package in general. You can also use virtualenv as a workspace when you can't copy files into site-packages because it's on a shared host. When you create a virtual environment with virtualenv, it creates a folder and copies Python into it along with a site-packages folder and a couple others. It also installs pip. Once your virtual environment is active, it's just like using your normal Python. And when you're done, you can just delete the folder to cleanup. No muss, no fuss. Alternatively, you can keep on using it for development.

In this article, we'll spend some time getting to know virtualenv and how to use it to make our own magic.

Installation

First of all, you probably need to install virtualenv. You can use pip or easy_install to install it or you can download the virtualenv.py file from their website and install it that way. At this point, assuming your Python folder is on the system path, you should be able to call virtualenv on the command line

Creating a Virtual Environment

Creating a virtual sandbox with the virtualenv package is quite easy. All you need to do is the following:

python virtualenv.py FOLDER_NAME

Where FOLDER_NAME is the name of the folder that you want your sandbox to go. On my Windows 7 machine, I have C:\Python26\Scripts added to my path so I can just call virtualenv.py FOLDER_NAME without the python part. If you don't pass it anything, then you'll get a list of options printed out on your screen. Let's say we create a project called sandbox. How do we use it? Well, we need to activate it. Here's how:

On Posix you would do source bin/activate while on Windows, you would do \path\to\env\Scripts\activate on the command line. Let's actually go through these steps. We'll create the sandbox folder on our desktop so you can see an example. Here's what it looks like on my machine:

C:\Users\mdriscoll\Desktop>virtualenv sandbox
New python executable in sandbox\Scripts\python.exe
Installing setuptools................done.
Installing pip...................done.

C:\Users\mdriscoll\Desktop>sandbox\Scripts\activate
(sandbox) C:\Users\mdriscoll\Desktop>

You'll note that once your virtual environment is activated, you'll see your prompt change to include a prefix of the folder name that you created, which is "sandbox" in this case. This lets you know that you're using your sandbox. Now you can use pip to install other packages to your virtual environment. When you're done, you just call the deactivate script to exit the environment.

There are a couple of flags you can pass to virtualenv when creating your virtual playground that you should be aware of. For example, you can use --system-site-packages to inherit the packages from your default Python's site packages. If you want to use distribute rather than setuptools, you can pass virtualenv the --distribute flag.

virtualenv also provides a way for you to just install libraries but use the system Python itself to run them. According to the documentation, you just create a special script to do it. You can read more here

There's also a neat (and experimental) flag called --relocatable that can be used to make the folder relocatable. However, this does NOT work on Windows at the time of this writing, so I wasn't able to test it out.

Finally, there's an --extra-search-dir flag that you can use to keep your virtual environment offline. Basically it allows you to add a directory to the search path for distributions that pip or easy_install can install from. In this way, you don't need to have access to the internet to install packages.

Wrapping Up

At this point, you should be able to use virtualenv yourself. There are a couple of other projects worth mentioning at this point. There's Doug Hellman's virtualenvwrapper library that makes it even easier to create, delete and manage virtual environments and then there's zc.buildout which is probably the closest thing to virtualenv that could be called a competitor. I recommend checking them both out as they might help you in your programming adventures.