Python by itself is awesome and packages add a lot of functionality. Installing all those packages makes a mess and some of them don’t play well with others, or maybe you need a certain version of a package for a project. Isolating is the way to go, virtualenv is one way to manage that.
One of the things I don’t like about using a Ubuntu LTS, rather than a distribution with rolling releases, is that many apt packages are locked in to the last stable version of the app. One day I thought I would do some work in python, I used pip to install a package, and pip dropped a message that I should update pip. So I checked for updates in apt-get, no new pip version there, so I followed the pip instruction to update through pip install --upgrade pip… then it broke. So much for being productive, lesson is don’t upgrade apt managed packages outside the apt package manager. The issue after doing an update that way is:
from pip import main
ImportError: cannot import name 'main'
Fixing this is easy once you find someone else’s solution, either remove and reinstall python3-pip or use pip as a python package rather than standalone command. https://stackoverflow.com/questions/49836676/error-after-upgrading-pip-cannot-import-name-main
Standalone pip
pip3 list
Vs. running pip through python
python3 -m pip list
Install this one global pip package, and no other’s if you can help it.
python3 -m pip install virtualenv
It should be runnable from command line after that
virtualenv --version
Choose where to setup your next project location and change directory to that location. It seems standard to name the virtualenv folder env or ENV, but it can be named something else.
virtualenv ~/Documents/python/newfolder/env
Each time you want to use a virtual environment, you will have to activate it for that session.
source ~/Documents/python/newfolder/env/bin/activate
We can confirm the paths are set for this virtualenv.
$ which pip
.../Documents/python/newfolder/env/bin/pip
$ which python
.../Documents/python/newfolder/env/bin/python
Installing packages will put the packages into the activated virtualenv, no mess, or version problems. As an example:
pip install scipy numpy pandas
When done with the virtualenv close the shell or deactivate the virtualenv.
deactivate
Check the docs https://pypi.org/project/virtualenv/ There’s options to set a specific python to use, like if you want set the virtualenv to use python2 or python3, or make the virtualenv project folder relocatable.