pytest.mark.xfail from inside the test is ignored

[tsyg]

When pytest.mark.xfail is used inside a test - it is not causing the test to be marked as xfail.
The sample code:

import pytest

@pytest.mark.parametrize("way", [
    "fresh",  # OK
    pytest.param("mix", marks=pytest.mark.xfail(reason="known_issue_with_MIX", strict=True)),
    "mixed",
])
def test_demo(way):
    print(f"testing - {way}")
    if way.startswith("mix"):
        pytest.mark.xfail(reason="mix has known issue!")  # Here - is ignored?
        print("NOTE: Marked sa xfail - see the result..")
    assert way.startswith("fr"), "only 'fr*' way is OK"

Expected result:
"fresh" test passed, "mix" and "mixed" ones treated as known issue (xfail)

Actual result: "mixed" test is shown failed.

> pip list
Package            Version
------------------ ----------
AllPairs           2.0.1
allpairspy         2.5.1
certifi            2023.11.17
charset-normalizer 3.3.2
colorama           0.4.6
idna               3.6
iniconfig          2.0.0
packaging          23.2
pip                23.3.1
pluggy             1.3.0
pyserial           3.5
pytest             7.4.3
requests           2.31.0
setuptools         69.0.2
urllib3            2.1.0
wheel              0.41.3

got the same in another venv:
pytest: 8.1.1
Python: 3.12.2
Platform: Windows-11-10.0.22631-SP0
pluggy: 1.5.0
Plugins:

• html: 4.1.1
• metadata: 3.1.1

[RonnyPfannschmidt]

creating a mark object without pitting it onto a test node is not going to do anything, did you mean to invoke pytest.xfail?

for applying the xfail mark later, its necessary to apply it to the current test item

[tsyg]

Thank you for the quick response.
Yes, I mean invoking pytest.xfail. I now found the sample in docs.
Once I replaced with pytest.xfail(reason="mix has known issue!") - Got the expected behavior.

Sorry for false alarm. Let's close the ticket.
Thanks!