Docker.github.io: docs: Could not find a version that satisfies the requirement cloud-init==20.4

[longqua69]

Hi Docker team,

In this training language/python/build-images.md, I have an issue with:

pip3 install -r requirements.txt

I've tried to investigate the root cause but I can't resolve this error now. I guess the ERROR caused by the parent image of this course: python:3.8-slim-buster

I'm trying to figure out what the root cause is. Maybe this ERROR is not caused by Docker but I think we should add some notes for watchers.

Thanks & Best regards,

Anh Tran

[Stig124]

I believe that you ran that on the system python install (and that MS has added some specific python packages in wsl)

Python uses virtual environnements (virtualenv) to isolate packages install, and permit to create independant packages

I suggest you to run a virtualenv using Pipenv and then continue the tutorial

[longqua69]

Hello @Stig124,

Thanks for your contact. I tried the way in your shared link Pipenv. But this way doesn't work.

The fact that, I don't really know your suggestion due to I don't have enough knowledge currently (I've not understood about virtual environment, dependencies manager... and how they work).

If I didn't understand and follow correctly your intention. Please let me know ^^!

Thanks,

Anh Tran

[Stig124]

I'll give you a requirements.txt when I'll be home

[longqua69]

Hi @Stig124,

Thanks for your dedicated support

[Stig124]

@longqua69

Paste that into the requirements.txt file

click==7.1.2
Flask==1.1.2
itsdangerous==1.1.0
Jinja2==2.11.3
MarkupSafe==1.1.1
Werkzeug==1.0.1

[longqua69]

Hi @Stig124,

I followed your guide. But it didn't work.

Do you have any idea?

[Stig124]

No, replacing the contents, not appending

Your whole file must contain only what I sent you

[longqua69]

Hi,

I got it. It works.

Thank you so much for your support.

Could you please share me what is the root cause? Do we need to add more notes in the document for helping other people?

Best regards,

Anh Tran

[Stig124]

When you're doing pip freeze > requirements.txt you are reporting every packages installed (so you can have offline modules that can't be ported outside of the local machine (I think that cloud-init is one of those)...), not the one that you just installed

The usage of a virtual environnement is to shield packages from a project to interfere with one another (some projects require specific versions of packages).

Take it like a Docker container, you have :

• the system pip (the host) where almost every piece of software (you still require root privileges to install packages on system pip) can look into to see what's installed,

• the user pip (rootless install of docker) which require a specific user to run it and is shielded from the system pip (packages are installed on home directory in opposition with the system-wide pip)

And then a

• virtualenv pip (a container), which is project-based, only that projects knows what is installed. The advantage of pipenv over other pieces of venv-wrapper (the venv is a built-in python component and critical for any developper that does python) are it's ease of use, you run pipenv install <package> and then the venv is created, with only one command. And it's got a requirements.txt-like file that even auto-updates

And yes, I really believe that a mention of using a virtualenv is a thing to add

[longqua69]

Hello @Stig124 ,

Thanks for your dedicated support. Now I can keep moving forward.

About your explanation, I haven't had enough knowledge to understand clearly what you've shared now. There are many terms that I've just known. But I can imagine the way it works because of your professional explanation.

For a clean, easy to self-study version of this document, could you please make a PR for closing this issue 😃? I think your explanation will be better than mine.

I'm waiting for your answer so I don't close this issue.

Best regards,

Anh Tran