Thanks for the catch@JohnVillalovos! We already have some scaffolding for mocking/prepping a clean config environment but perhaps it was not broad enough in where we placed it. It seems we should have a clean environment forall tests by adding that to the roottests/conftest.py e.g.:

1. Move thismock_clean_env fixture to the root conftest withautouse=True - no testever should break from a user-suppliedPYTHON_GITLAB_CFG:
   

   python-gitlab/tests/unit/test_config.py

   Lines 132 to 134 inf0152dc

   	@pytest.fixture
   	defmock_clean_env(monkeypatch):
   	monkeypatch.delenv("PYTHON_GITLAB_CFG",raising=False)

2. Move more logic to that fixture that we already use to ensure an empty list of default files - same here, no testever should be affected by presence of local files:
   

   python-gitlab/tests/unit/test_config.py

   Lines 144 to 145 inf0152dc

   	withmonkeypatch.context()asm:
   	m.setattr(config,"_DEFAULT_FILES", [])

3. In any test that does specifically need that, override it again if needed (I think we mostly don't, as we explicitly define things) like we already do here:
   

   python-gitlab/tests/unit/test_config.py

   Line 138 inf0152dc

   monkeypatch.setenv("PYTHON_GITLAB_CFG","/some/path")

More approaches are discussed herehttps://stackoverflow.com/questions/38748257/disable-autouse-fixtures-on-specific-pytest-marks in case the simple override is not enough.

WDYT? This would then also be more local/CI-agnostic and consistent across all runs and we can stop worrying about config even before we hit CI.