I agree, pathlib is really powerful, and I can see myself using it for new developments.
May I then assume that Refurb is better suited to work on new code? To get back to open(), I don’t know if I would want to go through my company’s legacy code (which runs on Python 3.10) and start replacing battle-tested code everywhere. Same would go for tuple literals over lists in certain circumstances.
I think this will go really well with choosing an „acceptable“ subset of Python for your project, similar to how some companies choose a subset of C++ and then stick with that throughout a project.
I think that using Refurb on new projects would be a great use case, but even for existing/old/legacy projects, Refurb can still be a good option, it really just depends on what is already there.
I agree that going and updating your company's codebase would probably be a bad idea, considering that there are some kinks here and there. Refurb tries its best, but it is very early on in its development, so errors that are emitted should be taken as suggestions, not gospel.
May I then assume that Refurb is better suited to work on new code? To get back to open(), I don’t know if I would want to go through my company’s legacy code (which runs on Python 3.10) and start replacing battle-tested code everywhere. Same would go for tuple literals over lists in certain circumstances.
I think this will go really well with choosing an „acceptable“ subset of Python for your project, similar to how some companies choose a subset of C++ and then stick with that throughout a project.