[REPO BUG] black vs flake linting might not be aligned.

[mi-volodin]

I have this issue for the following peace of code

class TestMatchedAndNotMatchedConditions(IncrementalBase):
    @pytest.fixture(scope="class")
    def seeds(self):
        return {
            "matched_and_not_matched_conditions_expected.csv": fixtures.matched_and_not_matched_conditions_expected,
        }

    @pytest.fixture(scope="class")
    def models(self):
        return {
            "matched_and_not_matched_conditions.sql": fixtures.matched_and_not_matched_conditions_model,
        }

    def test_merge(self, project):
        self.seed_and_run_twice()
        util.check_relations_equal(
            project.adapter,
            ["matched_and_not_matched_conditions", "matched_and_not_matched_conditions_expected"],
        )

First run with tox -e linter fails on flake8

Example output:

tests/functional/adapter/incremental/test_incremental_strategies.py:268:101: E501 line too long (116 > 100 characters)
tests/functional/adapter/incremental/test_incremental_strategies.py:274:101: E501 line too long (104 > 100 characters)
linter: exit 1 (0.65 seconds) /Users/dvolodin/Documents/GIT/dbt-databricks> .tox/linter/bin/python -m flake8 --select=E,W,F --ignore=E203,W503 --max-line-length=100 dbt tests pid=97727
  linter: FAIL code 1 (0.94=setup[0.04]+cmd[0.25,0.65] seconds)
  evaluation failed :( (1.03 seconds)

If I fix to something like this:

    @pytest.fixture(scope="class")
    def models(self):
        return {
            "matched_and_not_matched_conditions.sql": 
                fixtures.matched_and_not_matched_conditions_model,
        }

Black complains, because in black rules are more important than line length requirements (doc)

linter: commands[0]> .tox/linter/bin/python -m black --config black.ini --check dbt tests
would reformat /Users/dvolodin/Documents/GIT/dbt-databricks/tests/functional/adapter/incremental/test_incremental_strategies.py

Oh no! 💥 💔 💥
1 file would be reformatted, 148 files would be left unchanged.
linter: exit 1 (0.42 seconds) /Users/dvolodin/Documents/GIT/dbt-databricks> .tox/linter/bin/python -m black --config black.ini --check dbt tests pid=92649
  linter: FAIL code 1 (0.47=setup[0.05]+cmd[0.42] seconds)
  evaluation failed :( (0.54 seconds)

So it is a deadlock.

I can circumvent this easily, so no need to prioritise the fix. However I thought it does make sense to highlight this issue. Probably it will be useful in some maintainer round discussion as an argument to try other toolset (like ruff).

[benc-db]

Thanks for raising this; I've definitely hit this friction before, and it just hasn't reached a high enough priority to fix. I'm not sure that flake is even pulling it's weight anymore. Happy to hear proposals for other linting tools/configs.