test-writing-patterns

Arrange-Act-Assert divides a test into three sections: set up test data and dependencies (Arrange), execute the behavior being tested (Act), then verify the expected outcome (Assert). Keep the Act step to a single action and make assertions specific.

A stub returns predetermined responses, a mock verifies interactions and call parameters, a fake is a simplified working implementation like an in-memory database, and a spy wraps a real object to record calls while delegating to real behavior.

Use parameterized tests when the same test logic applies to multiple inputs, such as edge cases, boundary values, or validation rules. pytest uses @pytest.mark.parametrize and Jest uses test.each to run each case independently.

Brittle tests usually assert on implementation details instead of observable behavior. Fix this by testing outcomes and public interfaces rather than internal state, and avoid over-mocking dependencies.

No. Coverage only shows which code is executed by tests, not whether behaviors, edge cases, or error paths are verified. Focus on meaningful tests for critical paths rather than chasing coverage percentages.

Order-dependent tests violate isolation. Give each test its own setup and teardown, use fresh fixtures per test, avoid shared mutable state, and use transactions or in-memory databases that reset between tests.