If you have something that needs to be run before then it's a dependency of your test. So you should make a function out of what you want to have done before. I would not make a test that depends on a different test. If you do that than you will create a fail chain.