Why is Spring Data still using null references as return values?

Question

This is more a suggestion for improvement than a question.

We all know about Tony Hoare's "billion dollar mistake" by inventing the null reference. Google advices in the wiki of the guava library to avoid using null.

I really appreciate the Spring Data project and we use Spring Data MongoDB in many projects. Is there a chance that you will replace all possible null reference return values by an Optional<T>? I think this would be a big improvement when using the Spring Data repositories abstraction.

I know that a lot of the interfaces would have to be changed, but the code changes are almost trivial, just wrap the return type in an Optional.of(returnValue).

Answer 1

Guava/JDK8 Optional<T> support has already been introduced with RC1 of Release Train Dijkstra. Please have a look at the spring-data-examples for java8 to see how it works.

interface CustomerRepository extends Repository<Customer, Long> {

  // CRUD method using Optional
  Optional<Customer> findOne(Long id);

  // Query method using Optional
  Optional<Customer> findByLastname(String lastname);
}

BTW: there's support for default methods as well.

Answer 2

I think, once Spring Data can drop the support for JDK versions older than JDK 8, it should be fairly easy to use the JDK 8 Optional. However, since this would change the whole API, you would have to release a new major version, along with upgrade paths etc. (the code change would be trivial, but not the consequences for the users of Spring Data).

Answer 3

With the latest and greatest versions of Spring Data the CrudRepository now actually returns `Optional