In reply to Robb.
Indeed, that sounds like a very targeted mutation test. That could be especially useful when working with really large source files. Focusing on a specific area by hand may be quicker than letting a mutation tool generate hundreds of mutants that might not be relevant to your refactor.
I also like your focus on what makes a test fail. When I’m teaching developers how to write tests I often tell them that a unit test is only useful when it fails because that’s when you know that it has prevented a bug from reaching Production. It’s vitally important to think about what will make the test fail, because that will determine what bugs will or won’t be caught. As you pointed out, that goes for whether they’re writing tests from scratch or refactoring old ones.
