MAINT: Test using ruff with minimal changes to defaults #127
Conversation
|
Note that this uses much fewer rules than scipp/scipp#3424 In particular, it does not use rules based on isort, bugbear, and bandit. So we still need to run those linters. |
urghh, for some reason I thought the default was running all the rules (it's not). Need to select ALL. |
| "PT", "PYI", "RUF", "UP", "W"] | ||
| ignore = [ | ||
| "E111", "E114", "E117", "D206", "D300", # conflict with ruff format | ||
| "UP007"] # FIXME: PEP 604 breaks esssans test suite |
There was a problem hiding this comment.
Alright the problem is that this rule only works for python 3.10+ which essans satisfies but in the test suite it uses a released version of sciline which is on py3.9.
So I end up with errors like E TypeError: 'type' object is not subscriptable locally.
There was a problem hiding this comment.
Why would this matter? We don't lint our dependencies, are we?
| "parts = (CleanSummedQ[SampleRun, Numerator], CleanSummedQ[SampleRun, Denominator])\n", | ||
| "iofqs = (IofQ[SampleRun], IofQ[BackgroundRun], BackgroundSubtractedIofQ)\n", | ||
| "keys = monitors + (MaskedData[SampleRun],) + parts + iofqs\n", | ||
| "keys = (*monitors, MaskedData[SampleRun], *parts, *iofqs)\n", |
There was a problem hiding this comment.
It's more efficient. Plus it avoids the awkward (x,) syntax.
|
|
||
|
|
||
| def write_dependencies(dependency_name: str, dependencies: List[str]) -> None: | ||
| def write_dependencies(dependency_name: str, dependencies: list[str]) -> None: |
There was a problem hiding this comment.
Why is this preferred...? Or is it only for consistent type-hinting...?
There was a problem hiding this comment.
Using generics is deprecated behaviour https://docs.astral.sh/ruff/rules/non-pep585-annotation/
|
I'll close this one, once scipp/scipp#3424 is merged in we should create one in the template and let the github action do the work. |
This adds an example ruff config which uses all the defaults with minimal config. It makes very little changes.