This is in regard to tests of numpy.random but also in general. Can you have to many tests? Is there a reason not to have a test for each distribution in numpy.random? Vincent