Test Doubles in Automated Testing

Test Doubles are pretend objects used in place of real collaborators like databases and third-party APIs to make tests faster and more controllable. They are primarily used in unit and integration tests, while E2E tests should generally use real objects except for third-party services where flakiness is a concern. The article follows Gerard Meszaros and Martin Fowler's nomenclature, defining key concepts like SUT, collaborators, indirect inputs/outputs, and observation points.