test-driven-development

To write a failing test before implementing functionality, author a deterministic test specifying desired behavior, validate the failure, apply minimal code to pass, and run the full suite to confirm success. This red-green-refactor cycle prevents regressions.

The best way to reproduce a bug with a test is using the prove-it pattern: author a failing test reproducing the reported behavior, validate the failure, apply the minimal fix, and ensure the full test suite remains green to prevent regressions.

Choosing between stubs, mocks, and fakes depends on your testing strategy and test scale. Guidance covers the test pyramid, sizing tests as small, medium, or large, and making practical choices between real implementations and test doubles.

Yes, you can use test-driven development for integration and end-to-end testing. The workflow applies across unit, integration, and end-to-end test scopes, requiring deterministic tests runnable locally or in CI to verify behavior and prevent regressions.

You should refactor code during the TDD cycle after writing a failing test and implementing the minimal code to pass. Refactoring safely at this stage improves code quality while maintaining green tests and preventing regressions.

Test-driven development works for adding edge-case handling by first authoring a failing test specifying the edge case, validating the failure, applying the minimal code to pass, and ensuring the full test suite succeeds to prevent regressions.