Is it possible to limit the size of a @OneToMany collection with Hibernate or JPA Annotations?

rodrigobartels picture rodrigobartels · Oct 12, 2014 · Viewed 13.1k times · Source

If I have the following mapped collection in a POJO using JPA/Hibernate Annotations:

@OneToMany(mappedBy = "columnName", fetch = FetchType.LAZY)
@OrderBy("aField")
@MapKeyJoinColumn(name = "id_fk")
@LazyCollection(value = LazyCollectionOption.EXTRA)
private Set<PojoClass> collection = new HashSet<>();

Is it possible to limit the size of the loaded collection to a specific number n or to set the page size to lazy load the first n elements of the collection without loading all the items or any other items from other instances of the class (like with @BatchSize)?

Notice that the question explicitly refers to POJO mappings, not making additional queries, using Criteria or any other way to programmatically limit the size in explicit queries.

Possible solutions I've been looking at are finding an implementation of the SQL TOP statement within Hibernate, hopefully as an annotation or create a custom CollectionPersister to modify the generated query (although this would require having the TOP statement implemented within the supported dialects).

Answer

Vlad Mihalcea picture Vlad Mihalcea · Oct 12, 2014

That's not possible. Hibernate has two options for you:

  • you load the children collection upon fetching the parent entity
  • you load the children collection lazily on first access

There's no middle ground. That's because Hibernate needs to manage the whole collection entity state transitions. If you were able to load subsets then a unidirectional bag wouldn't make much sense.

While you can use @Where or @Filter, these annotations are more useful for filtering out the content of the collection, not to restrict its size.

So, you must always remember that Hibernate collections are a feature, not a mandatory requirement. Queries are way more flexible and less limiting.

So, you have to use queries this time:

  • HQL/JPQL
  • Criteria
  • Native SQL