[BUG] v13.3.4 broke coloring

[ofek]

Describe the bug

Previously setting NO_COLOR to 1 would override forced coloring but now it does not. The use case is in CI we force color to get nice test output but inside testing at the beginning of setup we set that environment variable so we can properly assert output.

[github-actions]

Thank you for your issue. Give us a little time to review it.

PS. You might want to check the FAQ if you haven't done so already.

This is an automated reply, generated by FAQtory

[willmcgugan]

As far as I can tell NO_COLOR does takes precedence over FORCE_COLOR.

Can you give me an example where that doesn't work?

[webknjaz]

@willmcgugan I just hit the exact same problem. Specifically, this is happening with pip, which vendors v13.4.2, currently: https://github.com/pypa/pip/blob/2a0acb5/src/pip/_vendor/vendor.txt#L15-L17.

We have FORCE_COLOR: 1 set for the entire GitHub Actions CI/CD workflow. But in one specific step of one job, we call pip ... --no-color | jq and it breaks because pip produces colored output through Rich and I don't think it uses Rich API to disable colorization: https://github.com/pypa/pip/blob/2a0acb595c4b6394f1f3a4e7ef034ac2e3e8c17e/src/pip/_internal/commands/install.py#L384.

There's one other place where it initializes Console() passing no_color explicitly: https://github.com/pypa/pip/blob/2a0acb595c4b6394f1f3a4e7ef034ac2e3e8c17e/src/pip/_internal/utils/logging.py#L268.

This is how it crashes for me: https://github.com/aio-libs/yarl/actions/runs/6934150793/job/18862144526#step:6:24.

@ofek is it pip that breaks in your case?

cc @pradyunsg does this sound like a pip bug?

[webknjaz]

Here's the hack that worked for me: aio-libs/yarl@f202823 (using a separate unset FORCE_COLOR Bash command).

I also ended up filing an issue in pip: pypa/pip#12405.

[ofek]

No it was for our main developer tool. I just had to stop forcing color DataDog/integrations-core@c5ef1ef

[webknjaz]

@ofek sounds like you also had to use a hack?

[willmcgugan]

It's been fixed for a while in Rich. Pip will be using an older vendored version.

[webknjaz]

@willmcgugan do you have the exact commit? I could attempt testing it with a local pip checkout...

[webknjaz]

I posted in #2449 (comment) a summary of the dissonance between two toggle that is likely happening here…

[webknjaz]

@willmcgugan do you have any workarounds in mind that could allow doing UNFORCE_COLOR?