Pytest parameterizing of a fixture by a test without using indirect

Question

Why does this work? Is it future pytest functionality, as yet undocumented? Anyone using this as an alternative to indirect parameterization? Seems like a much nicer syntax. I think pytest spins up the missing fixture using factory _pytest.fixtures.FixtureDef but haven't fully explored.

python 3.7.3, pytest 7.0.0

import pytest


@pytest.fixture
def c(d, e):
    return d + e


@pytest.mark.parametrize(argnames='d, e', argvalues=[(0, 2), (1, 3)])
def test_1(c, d, e):
    assert c == d + e

Expected it to fail as fixture c uses fixtures d + e that are not defined.