Refine the codebase

[jaraco]

I've been doing this lately with ruff --select UP --fix. See also https://github.com/jaraco/skeleton/issues/112, where we're looking to apply this constraint across all projects.

[diogobaeder]

I'm all for automating the process, as long as we can have human reviews on the changes (e.g. PRs or review-before-push or so)

[bswck]

I don't see much to do here when it comes to Python 3.8 modernization precisely. I'll then go ahead and say I'll generically improve the code quality.

[diogobaeder]

An idea would be to make use of things that were introduced in Python 3.8 but weren't available in earlier versions, where it makes sense. For example, in 3.6 dataclasses was an external library, and since 3.8 (IIRC) they became part of the stdlib. So we could cut out any reference to dataclasses as a library.

Type hinting also improved, so we could make use of that too I guess. What do you think?

[bswck]

I'll start from type hints.

[bswck]

For example, in 3.6 dataclasses was an external library, and since 3.8 (IIRC) they became part of the stdlib. So we could cut out any reference to dataclasses as a library.

Yeah, dataclasses was added to stdlib in 3.7. But I can see it between other stdlib imports here: https://github.com/coherent-oss/python-kustomize/blob/a0c5b3c22a5cecd2704a47df87e0aeb00a094607/kustomize/generators.py#L6-L8

And I don't see a place where dataclasses is referred to as if it was a third-party package.

[diogobaeder]

Here's a few ideas to look for:

• https://docs.python.org/3/whatsnew/3.7.html#whatsnew37-pep563 - import future annotations and use where classes reference themselves
• https://docs.python.org/3/whatsnew/3.8.html#assignment-expressions - "walrus operator", where it makes sense
• https://docs.python.org/3/whatsnew/3.8.html#positional-only-parameters - positional-only parameters in method headers that could be confusing without them (if this change is implemented, it makes the library's API backwards-incompatible, so we need to make sure to release at least a minor version next time if this happens)
• https://docs.python.org/3/whatsnew/3.8.html#f-strings-support-for-self-documenting-expressions-and-debugging - use f-strings where it makes sense

Other than this, I would be happy if we can make sure the package is mypy-strict. For some years I've been trying hard to keep type hints as strict as possible, because they're immensely helpful.

[bswck]

Thanks for assistance. I've been doing such refactors from time to time (https://github.com/CERT-Polska/malduck/pull/111, https://github.com/python-poetry/cleo/pull/365), I'm sure we'll get as much as we can out of it. :)

[bswck]

docs.python.org/3/whatsnew/3.7.html#whatsnew37-pep563 - import future annotations and use where classes reference themselves

I'm very happy you don't oppose to PEP 563. Opinions tend to be pretty controversial.