Skip to main content
CSE 331: Software Design & Implementation

CSE 331: Testing

Code is correct if it meets its specification. There are three ways to gain knowledge that code is correct: testing, tools (such as type checkers), and reasoning. This lecture covers testing.

A test runs code on a chosen input and checks the actual behavior against the expected behavior. Writing a test takes four steps: choose an input, use the spec to figure out the expected behavior, run the code, and compare. It is essential to determine the expected behavior before running the code and observing the actual behavior. A test of one method on its own is a unit test.

Choosing the input is the hardest step. Four tips for choosing inputs: test only inputs allowed by the specification, test each method separately, test exhaustively when possible, and cover each case and each boundary.

A specification-based test checks only behavior that the specification promises, so it passes on every correct implementation. An implementation-based test also checks behavior the specification does not promise, so it can fail on a correct implementation if that implementation happens to make different choices. No set of tests confirms a claim about every run. In the next lecture, we will discuss how many tests are enough.