EclipseLink JPA 2.1

转自:http://wiki.eclipse.org/EclipseLink/Release/2.5/JPA21
This page contains a summary of the major features supported in EclipseLink that implements the JPA 2.1 (JSR 338) specification requirements. The features and examples on this page do not represent a complete list. For more information, please see: the JSR 338 page.

Bulk Update

Until JPA 2.1, performing deletes or updates was not available using the Criteria API. Through theaddition of CriteriaUpdate/CriteriaDelete classes, support for bulkupdate/delete queries has now been added.

Update Example

The following example will update the salary and status, of all Employees who make less than 10000$, and give them a raise.

  CriteriaUpdate q = cb.createCriteriaUpdate(Employee.class);
  Root emp = q.from(Employee.class);
  .set(e.get(Employee_.salary), cb.prod(e.get(Employee_.salary), 1.1f))
  .set(e.get(Employee_.status), "full_time")
  .where(cb.lt(emp.get(Emploee_.salary), 10000));

The following Java Persistence query language update statement is equivalent.

  UPDATE Employee e SET e.salary = e.salary * 1.1, e.status = "full_time" WHERE e.salary < 10000

Delete Example

The following example deletes all the PhoneNumbers that are no longer in service

  CriteriaDelete q = cb.createCriteriaDelete(PhoneNumber.class);
  Root p = q.from(PhoneNumber.class);
  q.where(cb.equal(p.get(PhoneNumber_.status), "out_of_service"),

The following Java Persistence query language delete statement is equivalent.

  DELETE FROM PhoneNumber p
    WHERE p.status = 'out_of_service'

Stored Procedures

JPA specification 2.1 has introduced support for executing Stored Procedure calls. This includes a newStoredProcedureQuery API and Named Stored Procedure Queries (pre-existing portions of code on the database).

All the stored procedure examples below assume stored procedures already exist on the DB. Stored procedure creation is performed differently on different Databases. All the following example Stored procedure creation is using MySQL syntax (unless otherwise specified).

Simple Result Set example

Stored procedure definition on MySQL:

   CREATE PROCEDURE getIds() BEGIN SELECT ID FROM CUSTOMER ORDER BY ID ASC; END !

Build the query:

   StoredProcedureQuery spq = em.createStoredProcedureQuery("getIds", Customer.class);

Execute the query:

   List customers = spq.getResultList();

Alternatively, users can call spq.execute() directly (which is what getResultList() will call behind the scenes). The execute method will return a boolean indicating true if a result set is returned and false otherwise.

   boolean result = spq.execute();
   if (result == true) {
       customers = spq.getResultList();
   } else {
       // Handle the false for no result set returned, e.g.
       throw new RuntimeException("No result set(s) returned from the stored procedure"); 
   }

In parameter Example

Stored procedure definition on MySQL:

   CREATE PROCEDURE Update_Address_Postal_Code (new_p_code_v VARCHAR(255), old_p_code_v VARCHAR(255))
     BEGIN 
       UPDATE ADDRESS SET P_CODE = new_p_code_v WHERE P_CODE = old_p_code_v; 
     END

Build the query:

   StoredProcedureQuery spq = em.createStoredProcedureQuery("Update_Address_Postal_Code");
   spq.registerStoredProcedureParameter("new_p_code_v", String.class, ParameterMode.IN);
   spq.registerStoredProcedureParameter("old_p_code_v", String.class, ParameterMode.IN);

Execute the query:

   spq.setParameter("new_p_code_v", "123 NEW");
   spq.setParameter("old_p_code_v", "321 OLD");
   int updateCount = spq.executeUpdate();

Alternatively, the user could call the execute method directly (also note the parameters can be chained):

   spq.setParameter("new_p_code_v", "123 NEW").setParameter("old_p_code_v", "321 OLD").execute();
   int updateCount = spq.getUpdateCount();

OUT parameter Example

Stored procedure definition on MySQL:

   CREATE PROCEDURE Read_Address_City (address_id_v INTEGER, OUT city_v VARCHAR(255))
     BEGIN 
       SELECT CITY INTO city_v FROM ADDRESS WHERE (ADDRESS_ID = address_id_v); 
     END

Build the query:

   StoredProcedureQuery query = em.createStoredProcedureQuery("Read_Address_City");
   query.registerStoredProcedureParameter("address_id_v", Integer.class, ParameterMode.IN);
   query.registerStoredProcedureParameter("city_v", String.class, ParameterMode.OUT);

Execute the query:

   boolean resultSet = query.setParameter("address_id_v", "1").execute();
   
   if (resultSet) {
       // Result sets must be processed first through getResultList() calls.
   } 
   
   // Once the result sets and update counts have been processed, output parameters are available for processing.
   String city = (String) query.getOutputParameterValue("city_v");

Ref cursor Example

Stored procedure definition on Oracle:

   CREATE PROCEDURE Read_Using_Sys_Cursor (f_name_v VARCHAR2, p_recordset OUT SYS_REFCURSOR) AS
     BEGIN 
       OPEN p_recordset FOR SELECT EMP_ID, F_NAME FROM EMPLOYEE WHERE F_NAME = f_name_v ORDER BY EMP_ID; 
     END;

Build the query:

   StoredProcedureQuery query = em.createStoredProcedureQuery("Read_Using_Sys_Cursor", Employee.class);
   query.registerStoredProcedureParameter("f_name_v", String.class, ParameterMode.IN);
   query.registerStoredProcedureParameter("p_recordset", void.class, ParameterMode.REF_CURSOR);

Execute the query:

   query.setParameter("f_name_v", "Fred");                
   boolean execute = query.execute();
   List employees = (List) query.getOutputParameterValue("p_recordset");

Named Stored Procedure Example

Named stored procedures are those that are specified through metadata and uniquely identified by name.

Stored procedure definition on MySQL:

   CREATE PROCEDURE Read_Address (address_id_v INTEGER)
     BEGIN 
       SELECT ADDRESS_ID, STREET, CITY, COUNTRY, PROVINCE, P_CODE FROM ADDRESS WHERE (ADDRESS_ID = address_id_v); 
     END

Annotation Example

   @NamedStoredProcedureQuery(
       name = "ReadAddressByID",
       resultClasses = Address.class,
       procedureName = "Read_Address",
       parameters = {
           @StoredProcedureParameter(mode=IN, name="address_id_v", type=Integer.class)
       }
   )
   public Address() {
       ....
   }

XML Example

   
       
       Address
   

Execution

   EntityManager em = createEntityManager();
   em.createNamedStoredProcedureQuery("ReadAddressByID").setParameter("address_id_v", 1).getSingleResult();

A more complex example with multiple result sets using sql result set mappings

Stored procedure definition on MySQL:

   CREATE PROCEDURE Read_Multiple_Result_Sets ()
     BEGIN 
       SELECT E.*, S.* FROM EMPLOYEE E, SALARY S WHERE E.EMP_ID = S.EMP_ID; 
       SELECT A.* FROM ADDRESS A; 
       SELECT (t1.BUDGET/t0.PROJ_ID) AS BUDGET_SUM, t0.PROJ_ID, t0.PROJ_TYPE, t0.PROJ_NAME, t0.DESCRIP, t0.LEADER_ID, t0.VERSION, t1.BUDGET, t2.PROJ_ID AS SMALL_ID,
        t2.PROJ_TYPE AS SMALL_DESCRIM, t2.PROJ_NAME AS SMALL_NAME, t2.DESCRIP AS SMALL_DESCRIPTION, t2.LEADER_ID AS SMALL_TEAMLEAD, t2.VERSION AS SMALL_VERSION FROM
        PROJECT t0, PROJECT t2, LPROJECT t1 WHERE t1.PROJ_ID = t0.PROJ_ID AND t2.PROJ_TYPE='S'; 
       SELECT t0.EMP_ID, t0.F_NAME, t0.L_NAME, COUNT(t2.DESCRIPTION) AS R_COUNT FROM EMPLOYEE t0, RESPONS t2, SALARY t1 WHERE ((t1.EMP_ID = t0.EMP_ID) AND (t2.EMP_ID 
        = t0.EMP_ID)) GROUP BY t0.EMP_ID, t0.F_NAME, t0.L_NAME; 
     END

Build the query:

This is one example (of many) on how to configure such a query. Queries and result set mappings can be defined solely in annotations or xml or a mix of both. All the metadata can be defined on a single class or split up across many.

   @NamedStoredProcedureQuery(
       name="ReadUsingMultipleResultSetMappings",
       procedureName="Read_Multiple_Result_Sets",
       resultSetMappings={"EmployeeResultSetMapping", "AddressResultSetMapping", "ProjectResultSetMapping", "EmployeeConstructorResultSetMapping"}
   )
   
   @SqlResultSetMappings({
       @SqlResultSetMapping(
           name = "EmployeeResultSetMapping",
           entities = {
               @EntityResult(entityClass = Employee.class)
           }
       ),
       @SqlResultSetMapping(
           name = "EmployeeConstructorResultSetMapping",
           classes = { 
               @ConstructorResult(
                   targetClass = EmployeeDetails.class,
                   columns = {
                       @ColumnResult(name="EMP_ID", type=Integer.class),
                       @ColumnResult(name="F_NAME", type=String.class),
                       @ColumnResult(name="L_NAME", type=String.class),
                       @ColumnResult(name="R_COUNT", type=Integer.class)
                   }
               )
           }
       )
   })
   public Employee(){
       ....
   }

   @SqlResultSetMapping(
       name = "ProjectResultSetMapping",
       columns = {
           @ColumnResult(name = "BUDGET_SUM")
       },
       entities = {
           @EntityResult(
               entityClass = Project.class
           ),
           @EntityResult(
               entityClass = SmallProject.class,
               fields = {
                   @FieldResult(name = "id", column = "SMALL_ID"),
                   @FieldResult(name = "name", column = "SMALL_NAME"),
                   @FieldResult(name = "description", column = "SMALL_DESCRIPTION"),
                   @FieldResult(name = "teamLeader", column = "SMALL_TEAMLEAD"),
                   @FieldResult(name = "version", column = "SMALL_VERSION")
               },
               discriminatorColumn="SMALL_DESCRIM"
           )
       }
   )
   public Project() {
       ....
   }

   @SqlResultSetMapping(
       name = "AddressResultSetMapping",
       entities = {
           @EntityResult(entityClass = Address.class)
       }
   )
   public Address() {
       ....
   }

Execute the query:

   StoredProcedureQuery spq = createEntityManager().createNamedStoredProcedureQuery("ReadUsingMultipleResultSetMappings");
           
   // Read the first result set mapping --> Employee
   List employeeResults = spq.getResultList();
   
   // Read second result set mapping --> Address
   assertTrue("Address results not available", spq .hasMoreResults());
   List addressResults = spq.getResultList();
           
   // Read third result set mapping --> Project
   assertTrue("Projects results not available", spq .hasMoreResults());
   List projectResults = spq.getResultList();
           
   // Read fourth result set mapping --> Employee Constructor Result
   assertTrue("Employee constructor results not available", spq .hasMoreResults());
   List employeeConstructorResults = spq.getResultList();
           
   // Verify there as no more results available
   assertFalse("More results available", spq.hasMoreResults());

JPQL function

The SQL spec and many databases have SQL functions that are not covered by the JPA specification. With the latest JPAspecification the ability to call generic SQL functions was added to the JPQL syntax. The function keyword may be used to invoke predefined functions or used defined functions.

   SELECT e
     FROM Employee e
     WHERE FUNCTION(‘isLongTermEmployee’, e.startDate)

CDI Entity Listeners

Entity Listeners now support the Contexts and Dependency Injection API (CDI) when used inside a Java EE container. This support allows entity listeners to use CDI to inject objects and also provides support for @PostConstruct and @PreDestroy method calls.

CDI Example

The following example shows how a SessionBean can be injected into an EntityListener

 public class LoggerEntityListener {
   
   @EJB
   protected LoggerBean logger;
   
   @PrePersist
   public void prePersist(Object object) {
       logger.log("prepersist", object);
   }
   
   @PostPersist
   public void postPersist(Object object){
       logger.log("postpersist", object);
   }
   
   @PreDestroy
   public void preDestroy(){
       logger.close();
   }
   
   @PostConstruct
   public void postConstruct(){
       logger.initialize();
   }
 }

 @Entity
 @EntityListeners({LoggerEntityListener.class})
 public class MyLoggedEntity {
   ...
 }

Treat

Allows path expressions to be treated as a subclass, giving access to subclass specific state.

Treat JPQL Example

The following Java Persistence query language statement returns all Persons with a car that is a SportsCar with a maxSpeed of 200

  "Select p from Person p join fetch p.car join treat(p.car as SportsCar) s where s.maxSpeed = 200"

The following statement returns all maxSpeed values from referenced sportsCar instances

  Select s.maxSpeed from Person p join treat(p.car as SportsCar) s

Treat Criteria API Example

The following is equivalent to the first JPQL example above, returning Persions with a sportsCar having a maxSpeed of 200.

  CriteriaBuilder qb = em.getCriteriaBuilder();
  CriteriaQuery cq = qb.createQuery(Person.class);
  Root root = cq.from(Person.class);
  root.fetch("car");
  Join s = qb.treat(root.join("car"), SportsCar.class);
  cq.where(qb.equal(s.get("maxSpeed"), 200));

The following is equivelent to the second JPQL example, returning all maxSpeed values from referenced sportsCar instances

  CriteriaBuilder qb = em.getCriteriaBuilder();
  CriteriaQuery cq = qb.createQuery(Person.class);
  Root root = cq.from(Person.class);
  Join s = qb.treat(root.join("car"), SportsCar.class);
  cq.select(s.get("maxSpeed"));

Converters

Provides control over the conversion from an attribute type and value to the corresponding database type and value

Users must first define a class to the a converter. To do so, the class must implement the javax.persistence.AttributeConverter interface.

   public interface AttributeConverter {
   
       /**
        * Converts the value stored in the entity attribute into the 
        * data representation to be stored in the database.
        *
        * @param attribute  the entity attribute value to be converted
        * @return  the converted data to be stored in the database 
        *          column
        */
       public Y convertToDatabaseColumn (X attribute);
   
       /**
        * Converts the data stored in the database column into the 
        * value to be stored in the entity attribute.
        * Note that it is the responsibility of the converter writer to
        * specify the correct dbData type for the corresponding 
        * column for use by the JDBC driver: i.e., persistence providers are 
        * not expected to do such type conversion.
        *
        * @param dbData  the data from the database column to be 
        *                converted
        * @return  the converted value to be stored in the entity 
        *          attribute
        */
       public X convertToEntityAttribute (Y dbData);
   }

The class must be then marked as a Converter class through an annotation or xml declaration.

A Converter annotation example

   @Converter(autoApply=true)
   public class LongToStringConverter implements AttributeConverter {
   
       @Override
       public String convertToDatabaseColumn(Long attribute) {
           return (attribute == null) ? null : attribute.toString(); 
       }
   
       @Override
       public Long convertToEntityAttribute(String dbData) {
           return (dbData == null) ? null : new Long(dbData);
       }
   }

A Converter xml example

Note the auto apply flag set to true. This flag indicates that this converter should be applied to every attribute within the persistence unit of type Long. Without this setting, converters must be explicitly set with a Convert specification. E.g.

A Basic annotation example

   @Convert(converter = LongToStringConverter.class)
   protected Long salary;

A Basic xml example

Alternatively, an auto apply setting can also be turned using the Convert metadata for individuals attributes that wish to not use the Converter.

Annotation example:

   @Convert(disableConversion = true)
   protected Long salary;

XML example:

More complex usage

Converters can be applied at several levels. The simplest, described above, allows a single converter on a single attribute. Multiple converters can be applied through the use of Converts metadata which is of value for embedded attributes and when converters are needed on a mapped superclass on a per entity basis.

Converts and Embedded

   @Embedded
   @Converts({
       @Convert(attributeName = "level", converter = LevelConverter.class),
       @Convert(attributeName = "health", converter = HealthConverter.class),
       @Convert(attributeName = "status.runningStatus", converter = RunningStatusConverter.class)
   })
   protected RunnerInfo info;

Notice the multiple convert for different attributes of the embedded. The dot notation also allows you to traverse nested embedded mappings as needed.

Converts and Embedded key map

Converters can also be applied to Embeddable keys of "to-many" mappings. For example:

   @OneToMany(mappedBy="race")
   @Converts({
       // Add this convert to avoid the auto apply setting to a Long.
       @Convert(attributeName="key.uniqueIdentifier", disableConversion=true),
       @Convert(attributeName="key.description", converter=ResponsibilityConverter.class)
   })  
   protected Map organizers;
   
   @Embeddable
   public class Responsibility {
       public Long uniqueIdentifier;
       public String description;
       ...
   }

Converts and ElementCollection

Converters may also be applied to element collections (key and value) as needed. For example:

   @ElementCollection
   @Converts({
       @Convert(attributeName="key", converter = DistanceConverter.class),
       @Convert(converter = TimeConverter.class)
   })
   protected Map personalBests;

Here the DistanceConverter is applied to the keys of the personalBests map and the TimeConverter to the values of the map.

Converts and Entity

Converters may also be applied at the class level to connect converters to attributes of a mapped superclass from a subclass Entity. For example:

   @Entity
   @Converts({
       @Convert(attributeName = "accomplishments.key", converter = AccomplishmentConverter.class),
       @Convert(attributeName = "accomplishments", converter = DateConverter.class),
       @Convert(attributeName = "age", converter = AgeConverter.class)
   })
   public class Runner extends Athlete {
       ...
   }
   
   @MappedSuperclass
   public class Athlete {
       protected Integer age;
   
       @ElementCollection
       // Sub class (Runner) will add convert to both key and value
       protected Map accomplishments;
   
       ...
   }

DDL generation

In previous versions for JPA, although DDL generation was present it was not standardized or required. JPA 2.1 has addedstandardized provider DDL generation and made DDL generation a requirement.

A summary of the enabling DDL properties are as follows:

• javax.persistence.schema-generation.database.action
  • specifies the action to be taken by the persistence provider with regard to the database artifacts
  • Valid values for this property: none, create, drop-and-create, drop
• javax.persistence.schema-generation.scripts.action
  • specifies which scripts are to be generated by the persistence provider
  • Valid values for this property: none, create, drop-and-create, drop
• javax.persistence.schema-generation.create-source
  • specifies whether the creation of database artifacts is to occur on the basis of the object/relational mapping metadata, DDL script, or a combination of the two.
  • Valid values for this property: metadata, script, metadata-then-script, script-then-metadata
• javax.persistence.schema-generation.drop-source
  • specifies whether the dropping of database artifacts is to occur on the basis of the object/relational mapping metadata, DDL script, or a combination of the two.
  • Valid values for this property: metadata, script, metadata-then-script, script-then-metadata
• javax.persistence.schema-generation.create-database-schemas
  • specifies whether the persistence provider is to create the database schema(s) in addition to creating database objects such as tables, sequences, constraints, etc.
• javax.persistence.schema-generation.scripts.create-target
  • specifies a java.IO.Writer configured for use by the persistence provider for output of the DDL script or a string specifying the file URL for the DDL script.
• javax.persistence.schema-generation.scripts.drop-target
  • specifies a java.IO.Writer configured for use by the persistence provider for output of the DDL script or a string specifying the file URL for the DDL script.
• javax.persistence.database-product-name
  • specified if scripts are to be generated by the persistence provider and a connection to the target database is not supplied.
  • The value of this property should be the value returned for the target database by the JDBC DatabaseMetaData method getDatabaseProductName.
• javax.persistence.database-major-version
  • specified if sufficient database version information is not included from the JDBC DatabaseMetaData method getDatabaseProductName.
  • This value of this property should contain the value returned by the JDBC getDatabaseMajor-Version method.
• javax.persistence.database-minor-version
  • specified if sufficient database version information is not included from the JDBC DatabaseMetaData method getDatabaseProductName.
  • This value of this property should contain the value returned by the JDBC getDatabaseMinor-Version method.
• javax.persistence.schema-generation.create-script-source
  • specifies a java.IO.Reader configured for reading of the DDL script or a string designating a file URL for the DDL script.
• javax.persistence.schema-generation.drop-script-source
  • specifies a java.IO.Reader configured for reading of the DDL script or a string designating a file URL for the DDL script.
• javax.persistence.schema-generation.connection
  • specifies the JDBC connection to be used for schema generation. This is intended for use in Java EE environments, where the platform provider may want to control the database privileges that are available to the persistence provider.
• javax.persistence.sql-load-script-source
  • specifies a java.IO.Reader configured for reading of the SQL load script for database initialization or a string designating a file URL for the script.

Usage examples

Any combination of those properties can be passed through a Persistence.generateSchema() call, through a Persistence.createEntityManagerFactory() call or directly in the persistence unit definition in the persistence.xml.

Persistence.xml

   
       org.eclipse.persistence.jpa.PersistenceProvider
       META-INF/advanced-ddl-orm.xml
       true
       
       
           
           
           
           
           
       
   

Persistence.generateSchema()

   Map properties = new HashMap();
   properties.put("javax.persistence.database-product-name", "Oracle");
   properties.put("javax.persistence.database-major-version", 12);
   properties.put("javax.persistence.database-minor-version", 1);
   properties.put("javax.persistence.schema-generation.scripts.action", "drop-and-create");
   properties.put("javax.persistence.schema-generation.scripts.drop-target", "jpa21-generate-schema-no-connection-drop.jdbc");
   properties.put("javax.persistence.schema-generation.scripts.create-target", "jpa21-generate-schema-no-connection-create.jdbc");
            
   Persistence.generateSchema("default", properties);

NOTE: All the JPA DDL persistence unit properties are available statically from org.eclipse.persistence.config.PersistenceUnitProperties.SCHEMA_GENERATION*

Persistence.createEntityManagerFactory()

   Map properties = new HashMap();
   properties.put(PersistenceUnitProperties.SCHEMA_GENERATION_DATABASE_ACTION, PersistenceUnitProperties.SCHEMA_GENERATION_DROP_AND_CREATE_ACTION);
   properties.put(PersistenceUnitProperties.SCHEMA_GENERATION_CREATE_SOURCE, PersistenceUnitProperties.SCHEMA_GENERATION_SCRIPT_SOURCE);
   properties.put(PersistenceUnitProperties.SCHEMA_GENERATION_CREATE_SCRIPT_SOURCE, new FileReader(new File(createSource)));
   properties.put(PersistenceUnitProperties.SCHEMA_GENERATION_DROP_SOURCE, PersistenceUnitProperties.SCHEMA_GENERATION_SCRIPT_SOURCE);
   properties.put(PersistenceUnitProperties.SCHEMA_GENERATION_DROP_SCRIPT_SOURCE, new FileReader(new File(dropSource)));
           
   Persistence.createEntityManagerFactory("default", properties);

Add Named Query

Dynamic queries can now be added to the Peristence Unit through the EntityManagerFactory.addNamedQuery(String name, Query query) as named queries and can be retrieved through EntityManager.createNamedQuery(...). Configuration elements of the query like query hints, flush mode, lock mode, etc. are retained in the named query as configured at the point of adding the named query but parameter values are not retained. If a named query of the same name is already registered it will be replaced by the newly added query.Once retrieved any configuration changes to a named query will not be reflected in subsequent retrievals unless the named query is updated through addNamedQuery(...)

   {
   Query query = em.createQuery("Select e from Employee e where e.firstName = :p1 order by e.id");
   query.setParameter("p1", name);
   query.setMaxResult(15);
   factory.addNamedQuery("Select_Employee_by_first_name", query);
   
   em = factory.createEntityManager();
   Query namedQuery = em.createNamedQuery("Select_Employee_by_first_name");
   assertFalse(namedQuery.isBound(namedQuery.getParameter("p1"))); // <- parameter value not retained
   assertTrue(namedQuery.getMaxResult() == 15);// <- max result configuration retained
   
   namedQuery.setMaxResult(10);
   //configuration updates to retrieved queries are not reflected in the named query unless the new config is re-added
   Query namedQuery = em.createNamedQuery("Select_Employee_by_first_name");
   assertTrue(namedQuery.getMaxResult() == 15);
   }

Entity Graphs

Entity graphs are a means to specify the structure of a graph of entities using entity model metadata. This entity graph consists of representations of attributes and in the case of multi-node entity graphs additional entity graphs to represent the related entities. An entity graph can be specified through annotations:

   @NamedEntityGraph(
       name="ExecutiveProjects",
       attributeNodes={
           @NamedAttributeNode("address"),
           @NamedAttributeNode(value="projects", subgraph="projects")
       },
       subgraphs={
           @NamedSubgraph(
               name="projects",
               attributeNodes={@NamedAttributeNode("properties")}
           ),
           @NamedSubgraph(
               name="projects",
               type=LargeProject.class,
               attributeNodes={@NamedAttributeNode("executive")}
           )
       }
   )

and later retrieved by name:

   EntityGraph employeeGraph = em.getEntityGraph("ExecutiveProjects");

Entity graphs can be created dynamically from scratch:

   EntityGraph employeeGraph = em.createEntityGraph(Employee.class);
   employeeGraph.addAttributeNodes("address");
   employeeGraph.addSubgraph("projects").addAttributeNodes("properties");
   employeeGraph.addSubgraph("projects", LargeProject.class).addAttributeNodes("executive");

or created from an existing named entity graph:

   EntityGraph employeeGraph = em.createEntityGraph("ExecutiveProjects");
   employeeGraph.addSubgraph("period").addAttributeNodes("startDate");

Once constructed or retrieved the entity graphs can then be used as templates for certain EntityManager operations like load and fetch. For instance applying the entity graph as a fetch graph through a query hint will cause EclipseLink to only load those attributes present in the entity graph and unlisted attributes would become fetchType=LAZY.

   EntityGraph employeeGraph = em.getEntityGraph("ExecutiveProjects");
   Employee result = (Employee) em.createQuery("Select e from Employee e").setHint("javax.persistence.fetchgraph", employeeGraph).getResultList().get(0);
   PersistenceUnitUtil util = em.getEntityManagerFactory().getPersistenceUnitUtil();
   assertFalse(util.isLoaded(result, "firstName"));
   assertFalse(util.isLoaded(result, "department"));
   assertTrue(util.isLoaded(result, "projects"));

The entity graph can also be used to force and entity subgraph to be loaded at query time with the query hint "javax.persistence.loadgraph" . When a load graph is applied all listed attributes will be loaded by the query and any unlisted attributes will be loaded based on their mapping fetchType settings.

   EntityGraph employeeGraph = em.getEntityGraph("ExecutiveProjects");
   Employee result = (Employee) em.createQuery("Select e from Employee e").setHint("javax.persistence.loadgraph", employeeGraph).getResultList().get(0);
   PersistenceUnitUtil util = em.getEntityManagerFactory().getPersistenceUnitUtil();
   assertTrue(util.isLoaded(result, "firstName"));
   assertFalse(util.isLoaded(result, "department"));
   assertTrue(util.isLoaded(result, "projects"));

Unsynchronized Persistence Contexts

When working with JTA EntityManagers in previous versions of the Java Persistence API the entity manager and persistence context was automatically synchronized with the transaction and any changes to entities managed by the persistence context would be written to the database when the transaction was committed. With Java Persistence API 2.1 a new synchronization type @PersistenceContext(synchronization=SynchronizationType.UNSYNCHRONIZED) has been added to the persistence context to allow it to propagate along with the active JTA transaction but not be synchronized to it. For any unsynchronized persistence context the changes within the managed entities will not be written to the database unless the persistence context has been explicitly joined to the transaction by the application through the EntityManager.joinTransaction() call. This can be useful when an application may need access to transactional resources but does not wish to write to the database until some later point perhaps multiple transactions have completed.