Repositories act as an abstraction layer between the domain and infrastructure, separating business logic from data persistence concerns.
The first method involves replacing the backend client, like an S3 client, with a mock object to verify that the correct methods are called.
Instead of mocking the client library, you can intercept and mock the HTTP requests it makes to the backend service, like DynamoDB.
Asserting on exact SQL strings is a fragile approach because minor refactoring, like adding parameters, will break the tests.
Using a shared development or staging database provides a realistic environment but introduces significant challenges with state management and cleanup.
Testcontainers allows you to programmatically spin up ephemeral database instances in Docker, enabling realistic and isolated roundtrip testing.
A final summary compares the speed, complexity, and implementation coupling of client mocking, service mocking, and testing against real or temporary services.
The Q&A covers choosing mocking frameworks, balancing unit and integration tests, handling data migrations, and integrating repository tests into CI/CD pipelines.
30:55Alex Banul
43:07Golo Roden
47:22Christian Kühn
42:34Ryan Latta
26:02Vincent Hennig & Anupam Krishnamurthy
1:42:16Sascha Depold & Parinaz Roghany
56:49Jens Happe
25:36Eric Deandrea
Jobs that call for the skills explored in this talk.
CONTIAMO GMBH
Berlin, Germany
SYSKRON GmbH
Regensburg, Germany
