Jakarta Data and Jakarta Persistence by Gavin King

Jakarta Data is designed to work with both relational and non-relational data stores, offering a standardized API for data access

The specification provides type-safe query methods through annotation processing, catching errors at compile time rather than runtime

Repository interfaces in Jakarta Data are stateless by design, unlike traditional JPA EntityManager which maintains a persistence context

Jakarta Data implementations support CRUD operations and query methods without requiring custom implementations or boilerplate code

Queries can be defined using method names, annotations, or JDQL (Jakarta Data Query Language) - a subset of JPQL with restrictions

The specification supports pagination, sorting, and projection capabilities through standard interfaces and annotations

Type safety is achieved through annotation processors and static metamodels, providing compile-time validation of queries and entity references

Natural keys are preferred over synthetic keys for entities, particularly when implementing equals() and hashCode()

Jakarta Data is designed to be implementable across different datastores while maintaining consistent semantics

The specification aims to simplify data access without hiding the underlying persistence technology or introducing unnecessary abstractions

Repository operations are extensible, allowing implementations to add custom functionality while maintaining standard compliance