I have some pytest tests that run a test using a variety of different parameter combinations, like so:

class TestCases:
    d_prob = [0.1, 0.8]
    f_dens = [0.001, 0.1]
    r_scale = [True, False]
    i_corr = [True, False]

    # Cartesian product
    cart = itertools.product(d_prob, f_dens, r_scale, i_corr)

    @pytest.fixture(params=cart)
    def params(self, request):
        return request.param

    def test_mytest(self, params):
        d, f, r, i = params
        # Do stuff...

However, this can quickly blow out into a large number of tests and be pretty slow. So, I would like to have the behaviour change based on whether the slow mark has been set. I.e. if slow is set then use the full parameter matrix, otherwise use a reduced set that is faster but less comprehensive.

I’m not sure how to cleanly achieve this though. I am familiar with running tests or not by attaching the slow mark to them, but I have no idea if I can somehow change my parameter matrix based on such a mark.