Questions and answers
- How do we know what test to write if we have no code to test?
We reframe this thinking. Tests help us design a small section of code upfront. We decide what interface we want for this code and then capture these decisions in the AAA steps of a unit test. We write just enough code to make the test compile, and then just enough to make the test run and fail. At this point, we have an executable specification for our code to guide us as we go on to write the production code.
- Must we stick to one test class per production class?
No, and this is a common misunderstanding when using unit tests. The goal of each test is to specify and run a behavior. This behavior will be implemented in some way using code – functions, classes, objects, library calls, and the like – but this test in no way constrains how the behavior is implemented. Some unit tests test only one function. Some have one test per public method per class. Others...
