tdd-patterns

Choosing the right test double involves picking the simplest option—dummy, stub, spy, mock, or fake—to reduce brittleness and over-mocking risk. Selecting minimal doubles keeps your unit tests focused and resilient to changes in neighboring components.

The best way to start TDD on a non-trivial story is building a walking skeleton: the thinnest end-to-end production slice that proves your pipeline works. This establishes a safe foundation before you iteratively add real domain logic.

Double-loop TDD coordinates an outer acceptance-level loop with a faster inner unit-level loop. This sequencing preserves user value at the boundary while iteratively driving low-level design details until the acceptance criteria are fully satisfied.

Use property-based tests when validating invariant-driven logic where traditional example-based tests miss edge cases. This testing style automatically generates diverse inputs to verify mathematical properties or data contracts that must hold universally.

Safely modify legacy code by writing characterization tests and approval tests to capture existing behavior as a safety net. These specialized test styles snapshot current complex outputs, allowing you to refactor with confidence without altering functionality.

Use outside-in TDD when you know the interface and want to drive design from acceptance boundaries. Choose inside-out TDD when you need to build up from the core domain model before addressing outer behavior layers.