Casting fixture parameter to list at the beginning of parameter parsing.

[Sup3rGeo]

Closes #5946

• Target the master branch for bug fixes, documentation updates and trivial changes.
• (NON APPLICABLE) Target the features branch for new features, improvements, and removals/deprecations.
• (NON APPLICABLE) Include documentation when adding new features.
• Include new tests or update existing tests when applicable.

Unless your change is trivial or a small documentation fix (e.g., a typo or reword of a small section) please:

• Create a new changelog file in the changelog folder, with a name like <ISSUE NUMBER>.<TYPE>.rst. See changelog/README.rst for details.
• Add yourself to AUTHORS in alphabetical order;

[nicoddemus]

Looks great, thanks @Sup3rGeo!

[Sup3rGeo]

It seems that because we now always have numpy in the testing environments, the part that return False in this function never gets executed/tested:

pytest/src/_pytest/python_api.py

Lines 529 to 539 in 4f2abd7

	def _is_numpy_array(obj):
	"""
	Return true if the given object is a numpy array. Make a special effort to
	avoid importing numpy unless it's really necessary.
	"""
	import sys
	np = sys.modules.get("numpy")
	if np is not None:
	return isinstance(obj, np.ndarray)
	return False

What to do about it?

EDIT: I guess parametrize the approx tests with and without the numpy module? maybe by monkeypatching sys.modules.get to return None on those tests?

[nicoddemus]

TBH I don't think we need to bother testing that, adding # pragma: no cover to the return False should be enough.

[nicoddemus]

Thanks for the PR and reference! 👍

[blueyed]

@nicoddemus
Adding numpy to all builds make the numpy tox env obsolete, doesn't it?
I'd rather not have it installed always, but in case this is kept it should clean up the env, jobs, and conditional skips we have for it.

[nicoddemus]

Oh right, thanks for catching that.

Practically I think it makes sense to install an additional dependency in order to have less jobs, as an extra dependency is usually much quicker to install than starting up an entire new job, which potentionally might run all tests again.

I will follow up with a fix.

[blueyed]

Just started with it already myself.

[blueyed]

I wanted to keep numpy optional, but what you've said sounds different?
Keep in mind that numpy might be very slow to install without a wheel.

=> #5977

[nicoddemus]

You are right.

Let's remove numpy from extras and use importorskip on the test.

I have the diff ready:

diff --git a/setup.py b/setup.py
index be38857b0..dcf63f6fd 100644
--- a/setup.py
+++ b/setup.py
@@ -28,7 +28,6 @@ def main():
                 "mock",
                 "nose",
                 "requests",
-                "numpy",
                 "xmlschema",
             ]
         },
diff --git a/testing/python/fixtures.py b/testing/python/fixtures.py
index 4fcab0245..9ef7f0245 100644
--- a/testing/python/fixtures.py
+++ b/testing/python/fixtures.py
@@ -4190,6 +4190,7 @@ def test_indirect_fixture_does_not_break_scope(testdir):


 def test_fixture_parametrization_nparray(testdir):
+    pytest.importorskip("numpy")
     testdir.makepyfile(
         """
         from numpy import linspace