testing

The testing pyramid is a structured approach that emphasizes a large base of fast unit tests, fewer integration tests, and minimal end-to-end tests. This distribution optimizes for speed, maintainability, and cost by catching bugs early where they're cheapest to fix.

Use the AAA pattern: Arrange (set up test data and mocks), Act (execute the code), and Assert (verify results). This structure clarifies intent, isolates business logic from external dependencies, and makes tests predictable and fast under 100ms.

Mock external dependencies like APIs, databases, and third-party services to keep unit tests isolated and fast. Don't mock the code under test or core business logic; test those directly to catch real bugs in your application's behavior.

Coverage goals depend on context, but prioritize high coverage of business logic, boundary conditions, and error paths. Aim for meaningful coverage that catches real failures, not vanity metrics; track which scenarios matter most for your product.

Use test utilities to control time explicitly—mock timers, resolve promises in order, and avoid flaky sleeps. Structure async tests with clear await points and assertions that verify state after completion, not during execution.

Use integration tests to verify interactions between components—database queries, API calls, message queues—where mocking would hide real bugs. Keep them fewer and slower than unit tests, focused on workflows that matter to users.