Python Like A Pro: Managing Project Dependencies
Why Do You Need This Guide?
Suppose you’ve already mastered the fine art of installing multiple versions of Python on your computer using Pyenv. In that case, you’re probably keen to control your Python projects and the packages they use.
For a pro-developer, this means mastering some additional tools — tools designed to help you maintain a healthy degree of separation between your ‘system’ dependencies and your project dependencies. Without these tools, your system quickly becomes a tangled mess of incompatible packages that break your projects and disrupt your flow.
Before You Begin
This guide follows on where ‘Python Like A Pro: Installing Python’ left off. To complete it, you’ll need the
pyenv tool installed on your system and Python 3.8.5 and Python 2.7.18, both installed using the
pyenv install command as discussed in that earlier guide.
Check you have Pyenv and the Python versions required using the following command:
If successful, you’ll see output similar to that below, confirming that
pyenv is working and that Python 3.8.5 and 2.7.18 are available. If not successful, go back to the earlier guide.
* system (set by /home/ben/.pyenv/version) 2.7.18 3.8.5
Finally, if you’d prefer to watch a video on this guide, scroll down to the bottom of the page and hit ‘play.’
Managing Python Project Dependencies
As a rule of thumb, it’s best if each Python project you work on has its own set of dependencies, both on the version of Python it needs, but also on any dependencies you install via pip. This model is especially true if your work involves tens or even hundreds of Python projects — like a microservices project.
In the past, giving each Python project its independence was difficult. There was no built-in way to achieve isolation between all your Python projects. Fortunately, the Pyenv ‘VirtualEnv’ project can fix this issue, and it works with all Python versions including version 2.
Pyenv-virtualenv lets you:
- isolate Python projects using a workspace known as a ‘virtual environment.’
- link a Python project to a specific Python version.
- isolate the pip dependencies within a Python project.
- automatically switch to the correct virtual environment for the current project.
Let’s get ‘hands-on’ and take a closer look at
pyenv virtualenv in action.
Step 1: Create A Python 3.8.5 Project
Make a new folder and in this folder begin a new Python 3 project by adding a ready-made Python program file like so:
mkdir python-3-project cd python-3-project curl https://raw.githubusercontent.com/benwilcock/buildpacks-python-demo/master/web.py -o web.py
Now, still within the
python-3-project folder, create a new Python 3.8.5 virtual environment using
pyenv virtualenv <version> <project-name> like this:
pyenv virtualenv 3.8.5 py3
Then, activate the new
py3 virtual environment within the
python-3-project folder as follows:
pyenv local py3
At this point, your prompt may change slightly to confirm that the virtual environment is active. The folder will contain a new hidden file called
.python-version. This file contains the name of the virtual environment currently active for this folder — in this case,
# .python-version py3
Step 2: Add Some Package Dependencies
pip install Flask gunicorn
Now, when you list the currently installed pip packages with
pip list, the list displayed includes Flask and Gunicorn. The version of Gunicorn should be
20.x.x or higher because this was the first version designed exclusively for Python 3. Freeze the pip installed packages and their versions into a file with the following command:
pip freeze > requirements.txt
The contents of the
requirements.txt file should be similar to those shown below:
# requirements.txt click==7.1.2 Flask==1.1.2 gunicorn==20.0.4 itsdangerous==1.1.0 Jinja2==2.11.2 MarkupSafe==1.1.1 Werkzeug==1.0.1
Step 3: Test The Application
Use Gunicorn and Flask to run the
web.py application as follows:
gunicorn --bind=0.0.0.0:8080 web:app
When Gunicorn starts, it reports its version as
20.x.x like so:
[2020-08-19 10:37:08 +0100]  [INFO] Starting gunicorn 20.0.4
Now, point your browser to
http://localhost:8080 and you’ll be greeted with the legend “Hello, World!"
Step 4: Leave The Project Folder
Finally, leave the
py3 virtual environment will deactivate itself, and your regular command prompt will reappear. The command
pyenv version will confirm which version of Python is currently active now you’ve left the
# result of 'pyenv version' (in my case) system (set by /home/ben/.pyenv/version)
Step 5: Rinse And Repeat With Python 2.8.17
Repeat steps 1-4 above, but this time, create a folder called
python-2-project and use Python version 2.7.18 as the basis of your virtual environment. The updated commands for step 1 are as follows:
mkdir python-2-project cd python-2-project pyenv virtualenv 2.8.17 py2 pyenv local py2
In step 2, install the same packages via
pip, but this time notice how the version of Gunicorn that pip installs changes to
19.x.x. Version 19 of Gunicorn was the last version to support Python 2.
In step 3, run your app in Gunicorn as before, but this time, you’ll notice that it’s Gunicorn version
19.x.x that has started:
[2020-08-19 11:14:52 +0000]  [INFO] Starting gunicorn 19.10.0
To discover more of what
pyenv can do for you, check out the pyenv website or try
pyenv --help. To get help on a specific command in pyenv type
pyenv <command> --help.
If you liked this guide, you might find these others in our ‘Python Like A Pro’ series useful:
- Install Python Like A Pro!
- Managing Python Global Packages Like A Pro
- Build Python Docker Containers Like A Pro!
- Run Python In Production Like A Pro!
Here’s the video to accompany this guide: