Alexandre Decan

I've a small suggestion (that *could* break things, though) about `with_columns`. Currently, this method accepts a single parameter that is either an expression or a list of expressions (i.e., `df.with_columns([...,...

I'm not suggesting to go for a full (and only) kwargs-based api but for a combination of *args and **kwargs. In particular I think it would make more sense to...

Thanks for your answer! I understand the rationale. But in that case, I would be in favour, for consistency, not to allow the use of `**kwargs` and rather to allow...

Perhaps less ergonomic, but consistent with your choice for not supporting `*args`. Anyway, I prefer to have `**kwargs` rather than nothing :-p Apart from this, and going back to your...

Good news :-)

It doesn't really highlight that the data frame is returned as-is with new columns (ie there is no projection).

AFAIK actionlint is used to lint workflow files, not action files...

Because linting an Action is different than linting a workflow. The name of the tool is maybe a bit misleading, but its purpose is well defined on the homepage: _"[actionlint](https://github.com/rhysd/actionlint)...

Indeed :-) But remember you started your message with "I got a lot of false positives", my point was it's not surprising to have false positives if the tool is...

Thanks!