Fake Object vs Mock

Direct Answer

A fake is a small working implementation used in tests, such as a Map-backed repository that really stores and retrieves values. A Mockito mock is configured per test with stubs and can record interactions for verification. A fake is useful when state should behave consistently across several calls; a mock is useful when a test needs controlled replies or must check a boundary interaction.

Code Example

// Fake: real implementation using HashMap
class FakeUserRepository implements UserRepository {
    private final Map<Long, User> storage = new HashMap<>();
    private Long nextId = 1L;
    
    public User save(User user) {
        if (user.id() == null) {
            user = new User(nextId++, user.name(), user.email());
        }
        storage.put(user.id(), user); // Real storage logic
        return user;
    }
    
    public Optional<User> findById(Long id) {
        return Optional.ofNullable(storage.get(id)); // Real lookup
    }
}

// Mock: Mockito-generated hollow shell
@Test
void testWithMock() {
    var mockRepo = mock(UserRepository.class);
    var user = new User(1L, "Alice", "alice@example.com");
    when(mockRepo.save(any())).thenReturn(user); // Programmed response
    
    var result = mockRepo.save(user);
    verify(mockRepo).save(user); // Only verifies interaction, no real storage
}

Common Mistakes & Pitfalls

• Calling an alternative database a perfect fake: H2 is a real database with its own dialect; passing tests do not prove PostgreSQL or MySQL behavior.
• Copying production complexity: a fake should stay small and focused on the relevant contract.
• Letting a fake drift: run shared contract tests against the fake and real implementation where practical.
• Over-scripting a mock: a long sequence of stateful stubs may be clearer as a small fake.
• Verifying every call: verify only interactions that are observable requirements.