Virtual Environments

Virtual Environments allow you to have seperate boxes of packages for each project you work on.

Let's say Project X requires Ciphey 4.2.0, but project Y requires 5.10.0. To get around the fact they both need different versions, we use a virtual environment.

Essentially we create a container for both projects. That way we can install whatever we want without conflicts in the confinement of this container!

The default tool for Virtual Environments is rather bad, and no one really uses it anymore.

So we'll install a new one.

pip install virtualenvwrapper

Windows use https://pypi.org/project/virtualenvwrapper-win/ instead

Pay attention to the output in the terminal. We need to know the location of virtualenvwrapper.sh.

Now in our .bashrc or .zshrc (the file that runs everytime a terminal is opened) add these lines:

export WORKON_HOME=$HOME/.virtualenvs
export PROJECT_HOME=$HOME/projects
source /usr/local/bin/virtualenvwrapper.sh

Replacing the source /usr/local.... with the file location outputted in the terminal.

Now, source your .bashrc:

source .bashrc

Note: this is confusing and you may get this wrong the first time. If so, please read more about .bashrc via Google 😄

Now, make a new folder and call it "my-new-project" or something.

cd into that folder and run:

mkvirtualenv my-new-project

This automatically creates the virtual environment for you, and drops you into it.

Now we can pip install anything we want, and it will only be associated with this virtual environment.

We can leave our virtual environment with deactivate. We no longer have access to the package we installed in the virtual environment.

To go back, run workon.

$ workon
my-new-project

This tells us the name(s) of our virtual environments. And then:

workon my-new-project

Takes us back there.

And just like that, we have now fixed our packaging problem with this neat tool (and trust me when I say, this is easier than the default tool).

Now, remember how I said that Ciphey works on 3.7 and up, but what if we have another piece of software that only works on 3.6 and below? How would we use different versions of Python?

This is solved by....