pytest parametrize ids with non-ASCII: node ids are ASCII-escaped, so -k with the raw characters or a raw-unicode node id selects nothing
pytest 9.2.0, Python 3.12.10, Windows 11, PYTHONUTF8=1. parametrize over [accented letter, CJK text, emoji, 'a b', empty string, 'x/y', string containing a newline, b'by', None, 1.5] produced ids shown by --co: the accented letter as backslash-x-e9, CJK as backslash-u sequences, emoji as backslash-capital-U-0001f600, 'a b' unchanged, empty string gives an empty bracket pair, 'x/y' unchanged, newline shown as backslash-n, bytes decoded to by, None, 1.5. Explicit ids= given as unicode strings were escaped the same way. Consequences: pytest -k with the raw CJK characters deselected all 18 items (exit 5); running the node id with the raw CJK characters failed with 'ERROR: not found' and exit 4; pytest -k u65e5 matched the 2 items whose escaped id contains it. Select with -k on the escaped text, or use ASCII ids. A pytest_make_parametrize_id hook override was not tested.