Python docker，我应该通过 pip（在图像内）添加新库并从 pip 冻结创建 requirements.txt，还是手动编辑 requirements.txt？

Question

let's say I am working on a python project with docker. using python as the base image. Now for new requirements, I need to install another library eg pillow. How do I manage it?

Run (python-docker) image in interactive mood and use these commands, pip install pillow, pip freeze > requirements.txt. so it will update requirements.txt and also lock the library version.

Or do I just check the internet for the latest/required pillow library version and manually insert it in requirements.txt? Trying to see how developers approach this on docker.

Could not find anything appropriate on the internet. They just have everything required ready beforehand in requirements.txt!

Answer 1

You should pretty much never run any sort of "install" command in an interactive shell in a container. Anything you do in this context will get lost as soon as you exit the container. Similarly, copying files out of a container is inconvenient, validating any updated libraries work is inconvenient, and so on.

Instead I'd set up an ordinary Python virtual environment

python3 -m venv venv
. venv/bin/activate

Install your current package set there, install the new (or updated) package, and update the requirements file.

pip install -r requirements.txt
pip install pillow
pip freeze > requirements.txt

While you're still in the virtual environment, you can double-check that things still work and see what's exactly changed.

pytest
git diff requirements.txt
./manage.py runserver

If this all works, then you're ready to do an integration test with the updated image.

docker build -t my/image .
docker stop my-app
docker rm my-app
docker run -d --name my-app ... my/image

And if that works then you have a change you've validated in two places that you're ready to commit.

git add requirements.txt
git commit -m 'add pillow dependency'

deactivate