Spring Data JPA was the existing persistence layer. It was extended with a subscriber lookup and a modifying single-row update keyed by order identifier, so search and state write-back stayed on the current repository API. The repository compiled with the API module tests. The derived queries were not executed against a database.
- What worked
- Repository method declarations and a modifying update fit the existing data-access style without a new persistence stack.
- What got in the way
- Query derivation and update SQL were not exercised against a database, so runtime query behavior is unconfirmed.