test(conftest): support @pytest.mark.bashcomp(required_cmd=True)

[akinomyoga]

Separate commit 53060e3 in #546

[scop]

Needs to be run through black, feel free to merge after fixing. Highly recommended to use pre-commit locally to avoid roundtrips like this.

[akinomyoga]

Ah, OK. I've fixed it. I thought I have run pre-commit previously for #544, but it seems I have skipped it because #544 is still a draft PR.

[akinomyoga]

The test has passed, so I merge this PR.

[scop]

I thought I have run pre-commit previously for #544, but it seems I have skipped it because #544 is still a draft PR.

I suggest installing pre-commit locally with pre-commit install so there is never need to remember to run it, it will just happen automatically when one tries to commit. git commit -n can be used to bypass it if it causes problems e.g. with one's workflow (intermediate WIP commits that should not be linted yet or so).

[akinomyoga]

Yeah, I once installed the pre-commit hook after you have suggested it to me before. But after that, pre-commit interfered with git rebase -i and caused troubles. Since I frequently do git rebase, I have uninstalled it and decided to run it manually.

[scop]

FWIW I use pre-commit and rebase with and without -i a lot, and I don't think I've ever run into problems with that combo.

[akinomyoga]

Oh, yeah. I didn't remember the exact situation that I met with git rebase, so I had actually re-installed the pre-commit hook again after my previous comment. Then, I remembered that it was related to misconfigured mypy tests. In my environment, mypy produces the following error, which prevent me of making commits touching any *.py files:

mypy.....................................................................Failed
- hook id: mypy
- exit code: 1

No virtual environments found

When I first installed the hook, I first faced with this issue in git rebase where I tried to modify an older commit and there are some newer commits touching *.py after the modified commit. Since you have mentioned that there is the same issue in your environment with mypy in #526 (comment), I just gave up to try to fix this mypy issue. Do you have any idea on how to fix this problem?

[scop]

Our dev env setup instructions are inadequate, indeed a python virtualenv must be set up in the bash-completion working dir, then the above should work. Basically the commands in this run section need to be run locally, or replacing the first two with a functionally equivalent pyenv and pyenv-virtualenv setup (that's what I'm using), venv-run then picks the env up and things should work in principle. The other mypy related issue you mentioned is something different, this doesn't fix that one, but I don't think I'm hitting it every time.

[akinomyoga]

OK thank you for your instruction! It now works correctly.