-k filters on markers also

[lodagro]

The -k option takes also markers into account. Based in the docs, this is unexpected (at least for me).
Below is a little example, using test file test_k_m.py. Two tests, the first one has mymark in its id and the seconds one has a marker mymark.

import pytest

@pytest.mark.parametrize("foo", [
    pytest.param(0, id="some_id_with_mymark_in_it"),
    pytest.param(1, id="id1", marks=(pytest.mark.mymark))
])
def test_stuff(foo):
    print(f"{foo=}")

Collecting tests using -k mymark selects both tests. I expected only test_stuff[some_id_with_mymark_in_it] to be selected.

❯ pytest test_k_m.py -k "mymark" --collect-only
==================================== test session starts ====================================
platform linux -- Python 3.11.4, pytest-7.4.0, pluggy-1.2.0
rootdir: /lwork/OLT/wo904355/sandbox
plugins: anyio-3.7.1, forked-1.6.0, mock-3.11.1, xdist-3.3.1
collected 2 items                                                                           

<Module test_k_m.py>
  <Function test_stuff[some_id_with_mymark_in_it]>
  <Function test_stuff[id1]>
================================ 2 tests collected in 0.01s =================================

Collecting tests using -m mymark works as expected.

❯ pytest test_k_m.py -m "mymark" --collect-only
==================================== test session starts ====================================
platform linux -- Python 3.11.4, pytest-7.4.0, pluggy-1.2.0
rootdir: /lwork/OLT/wo904355/sandbox
plugins: anyio-3.7.1, forked-1.6.0, mock-3.11.1, xdist-3.3.1
collected 2 items / 1 deselected / 1 selected                                               

<Module test_k_m.py>
  <Function test_stuff[id1]>
======================== 1/2 tests collected (1 deselected) in 0.01s ========================

[RonnyPfannschmidt]

yes, this is due to the fact that markers started as keywords being expanded about 17-19 year ago