Before JPA 2.1

There were two types of queries:

• Named (static): these are declared using the @NamedQuery or @NamedQueries annotations.
• Dynamic: as the name suggests, this type of query is specified by the application logic at runtime

JPQL, Native SQL, as well as Stored Procedure based queries support the above-mentioned modes and each one of them has their own pros and cons.

JPA 2.1 Supports

The notion of a hybrid style query wherein you can:

• declare a query in the code (as a regular string)
• bind it to the EntityManagerFactory
• use it as a regular Named Query

......
  private static final String FIND_ASSIGNEE = "SELECT a.assignee FROM Approvals a WHERE a.id = :reqId";

  @PersistenceContext(unitName="AccessRequestDataStore")
  private EntityManager em;

  public void test(String reqId) {

    //create once and bind
    Query query = em.createQuery(FIND_ASSIGNEE);
    em.getEntityManagerFactory().addNamedQuery("findAssignee", query);

    //use it
    String assignee = em.createNamedQuery("findAssignee")
    .setParameter("reqId", reqId)
    .getSingleResult();
  }
......

Predominant Use Case

Useful when a query has the following properties:

• It needs to be executed repeatedly (this one is obvious), and
• Cannot be pre-determined (else @NamedQuery would have sufficed !); i.e. it depends on run-time information

Advantages

Although it incurs a (one-time) processing overhead, after that, it offers the same benefits as a regular named query (i.e. does not need to be processed by the JPA provider repeatedly). This is obviously not the case with dynamic JPQL queries, which need to be processed by the provider (converted to SQL) prior to execution

Further Reading

• JPA 2.1 specification
• JPA javadoc