I have been spending much of my coding time in Python recently. This site is built in Pelican. Many lines of Python have been written for work. I have even been poking at Google App Engine in what spare time is available. The only disappointment is that all of these have been in Python 2. I would prefer to be using Python 3. There is a little free time today, so I will set up a nice Python 3 workspace.
One of the interesting things about Python is how it handles personal workspaces. Popular tools in other languages, such as rbenv for Ruby and perlbrew for Perl, focus on a complete localized installation for any version you care to use. Python tools assume a system standard version, and focus on making a snapshot for your projects. That works sort of like Bundler. Once you have your snapshot loaded, you use pip to install the exact libraries needed by your projects. That works very much like Bundler.
The tool of choice for making virtual environments in Python 3 is pyvenv. pyvenv actually comes with the standard installation of Python 3.3 or greater. That is good news. Python 2’s virtualenv was not hard to install, but it was not available by default. You still had to install it.
There is already an excellent introduction to using pyvenv. That tells most of what you need to know.
my-project holds the default Python interpreter until you exit
that particular shell or activate a different virtual environment.
That is good enough to get started, but I often have several projects going. Each project gets its own virtual environment. Having my environment files just sitting there in my project folder bothers me, though.
I like things tidy, so I use virtualenvwrapper. virtualenvwrapper
creates a single folder to hold your virtual environments and provides
you with a couple of shell commands to manage those environments. It’s
easy if you’re only using a single system Python. Just follow their
installation instructions. After that,
are your friends.
What if I wanted to use both Python 2 and Python 3 virtual environments? Well, the virtual environments I already had set up for work function without any problem. Thank goodness. If I wanted to do new Python 2 work - well, that may be a good time to pull up a real virtual environment with a tool like Vagrant. I may come back to that later.