Test It Before You Ship It — Java API Testing Lab

09:00 – 10:15

Draw the pyramid for your own service

—what each test layer buys, cost of each layer, mapping your current suite

—Takeaway Your own suite mapped onto the pyramid — with the gaps named out loud, which is the plan for the two days.

10:30 – 11:45

Write unit tests that earn their keep

—JUnit 5 lifecycle, parameterised tests, assertions that explain failure

—Takeaway A failing test whose message alone tells you the bug — no debugger, no print statements.

12:00 – 13:00

Isolate with Mockito, without mocking everything

—mocks vs stubs vs fakes, verify used sparingly, the over-mocked smell

—Takeaway One over-mocked test rewritten until it tests behaviour instead of implementation — and survives a refactor to prove it.

14:00 – 15:15

Test the web slice with Spring Boot Test

—@WebMvcTest and MockMvc, validation and error contracts, security in slices

—Takeaway Controller tests that fail when a status code, error shape or validation rule quietly changes.

15:30 – 17:00

Test the data layer against the real thing

—@DataJpaTest, Testcontainers vs H2, migrations under test

—Takeaway The data layer tested on real PostgreSQL in a container — and the bug H2 would have hidden, demonstrated.

09:00 – 10:15

Boot the whole service in a test

—@SpringBootTest honestly costed, slice vs full context, context caching

—Takeaway A full-context integration test that earns its runtime — and the suite reorganised so only a few need to.

10:30 – 11:45

Drive the API from the outside with REST Assured

—given/when/then requests, asserting the contract, auth in outside-in tests

—Takeaway The API exercised exactly as a client would — with the error contract asserted, not assumed.

12:00 – 13:00

Automate scenarios with Karate

—feature files, data-driven scenarios, where Karate beats code

—Takeaway A readable scenario file a non-developer could review — running against the same API, in CI like everything else.

14:00 – 15:15

Make the suite one you can trust

—flaky tests to root cause, test data builders, what not to assert

—Takeaway A planted flaky test hunted to its cause — time, order and shared state all interrogated on the way.

15:30 – 17:00

Turn it into a gate and defend it

—the suite in GitHub Actions, coverage read honestly, review

—Takeaway The whole pyramid running as a CI gate that blocks a real planted regression — and the argument for each layer, made to the room.