Managing contents of requirements.txt for a Python virtual environment

Question

So I am creating a brand new Flask app from scratch. As all good developers do, my first step was to create a virtual environment.

The first thing I install in the virtual environment is Flask==0.11.1. Flask installs its following dependencies:

click==6.6

itsdangerous==0.24

Jinja2==2.8

MarkupSafe==0.23

Werkzeug==0.11.11

wheel==0.24.0

Now, I create a requirements.txt to ensure everyone cloning the repository has the same version of the libraries. However, my dilemma is this:

Do I mention each of the Flask dependencies in the requirements.txt along with the version numbers OR
Do I just mention the exact Flask version number in the requirements.txt and hope that when they do a pip install requirements.txt, Flask will take care of the dependency management and they will download the right versions of the dependent libraries

Take a look at the `pip-compile` command in [pip-tools](https://github.com/nvie/pip-tools/blob/master/README.md). — dirn, Sep 09 '16 at 12:04

score 34 · Answer 1 · edited Oct 31 '20 at 09:23

One good thing here is you are using virtualenv, which will make your task very easy.

Activate virtualenv ($source path_to_virtualenv/bin/activate)
Go to your project root directory
Get all the packages along with dependencies in requirements.txt
```
pip freeze > requirements.txt
```
You don't have to worry about anything else apart from making sure next person installs the requirements recursively by following command
```
pip install -r requirements.txt
```

score 6 · Answer 2 · answered Sep 09 '16 at 07:45

You can (from your active virtual environment) do the following

pip freeze > requirements.txt

which'll automatically take care of all libraries/modules available in your project.

The next developer would only have to issue:

pip install -r requirements.txt

score 6 · Accepted Answer · edited Sep 27 '17 at 13:01

Both approaches are valid and work. But there is a little difference. When you enter all the dependencies in the requirements.txt you will be able to pin the versions of them. If you leave them out, there might be a later update and if Flask has something like Werkzeug>=0.11 in its dependencies, you will get a newer version of Werkzeug installed.

So it comes down to updates vs. defined environment. Whatever suits you better.

score 1 · Answer 4 · answered Nov 01 '18 at 20:35

1

If you only want to see what packages you have installed then just do
pip freeze.

but if you want all these packages in your requirement.txt, then do
pip freeze > requirements.txt

answered Nov 01 '18 at 20:35

Deke

4,451
4
44
65

Managing contents of requirements.txt for a Python virtual environment

4 Answers4