HIBERNATE - Relational Persistence for Idiomatic Java

Our first persistent class is a simple JavaBean class with some properties:

package events;

import java.util.Date;

public class Event {
    private Long id;

    private String title;
    private Date date;

    public Event() {}

    public Long getId() {
        return id;
    }

    private void setId(Long id) {
        this.id = id;
    }

    public Date getDate() {
        return date;
    }

    public void setDate(Date date) {
        this.date = date;
    }

    public String getTitle() {
        return title;
    }

    public void setTitle(String title) {
        this.title = title;
    }
}

You can see that this class uses standard JavaBean naming conventions for property getter and setter methods, as well as private visibility for the fields. This is a recommended design - but not required. Hibernate can also access fields directly, the benefit of accessor methods is robustness for refactoring. The no-argument constructor is required to instantiate an object of this class through reflection.

The id property holds a unique identifier value for a particular event. All persistent entity classes (there are less important dependent classes as well) will need such an identifier property if we want to use the full feature set of Hibernate. In fact, most applications (esp. web applications) need to distinguish objects by identifier, so you should consider this a feature rather than a limitation. However, we usually don't manipulate the identity of an object, hence the setter method should be private. Only Hibernate will assign identifiers when an object is saved. You can see that Hibernate can access public, private, and protected accessor methods, as well as (public, private, protected) fields directly. The choice is up to you and you can match it to fit your application design.

The no-argument constructor is a requirement for all persistent classes; Hibernate has to create objects for you, using Java Reflection. The constructor can be private, however, package visibility is required for runtime proxy generation and efficient data retrieval without bytecode instrumentation.

Place this Java source file in a directory called src in the development folder, and in its correct package. The directory should now look like this:

.
+lib
  <Hibernate and third-party libraries>
+src
  +events
    Event.java

In the next step, we tell Hibernate about this persistent class.

Hibernate needs to know how to load and store objects of the persistent class. This is where the Hibernate mapping file comes into play. The mapping file tells Hibernate what table in the database it has to access, and what columns in that table it should use.

The basic structure of a mapping file looks like this:

<?xml version="1.0"?>
<!DOCTYPE hibernate-mapping PUBLIC
        "-//Hibernate/Hibernate Mapping DTD 3.0//EN"
        "http://hibernate.sourceforge.net/hibernate-mapping-3.0.dtd">

<hibernate-mapping>
[...]
</hibernate-mapping>

Note that the Hibernate DTD is very sophisticated. You can use it for auto-completion of XML mapping elements and attributes in your editor or IDE. You also should open up the DTD file in your text editor - it's the easiest way to get an overview of all elements and attributes and to see the defaults, as well as some comments. Note that Hibernate will not load the DTD file from the web, but first look it up from the classpath of the application. The DTD file is included in hibernate3.jar as well as in the src/ directory of the Hibernate distribution.

We will omit the DTD declaration in future examples to shorten the code. It is of course not optional.

Between the two hibernate-mapping tags, include a class element. All persistent entity classes (again, there might be dependent classes later on, which are not first-class entities) need such a mapping, to a table in the SQL database:

<hibernate-mapping>

    <class name="events.Event" table="EVENTS">

    </class>

</hibernate-mapping>

So far we told Hibernate how to persist and load object of class Event to the table EVENTS, each instance represented by a row in that table. Now we continue with a mapping of the unique identifier property to the tables primary key. In addition, as we don't want to care about handling this identifier, we configure Hibernate's identifier generation strategy for a surrogate primary key column:

<hibernate-mapping>

    <class name="events.Event" table="EVENTS">
        <id name="id" column="EVENT_ID">
            <generator class="native"/>
        </id>
    </class>

</hibernate-mapping>

The id element is the declaration of the identifer property, name="id" declares the name of the Java property - Hibernate will use the getter and setter methods to access the property. The column attribute tells Hibernate which column of the EVENTS table we use for this primary key. The nested generator element specifies the identifier generation strategy, in this case we used native, which picks the best strategy depending on the configured database (dialect). Hibernate supports database generated, globally unique, as well as application assigned identifiers (or any strategy you have written an extension for).

Finally we include declarations for the persistent properties of the class in the mapping file. By default, no properties of the class are considered persistent:

<hibernate-mapping>

    <class name="events.Event" table="EVENTS">
        <id name="id" column="EVENT_ID">
            <generator class="native"/>
        </id>
        <property name="date" type="timestamp" column="EVENT_DATE"/>
        <property name="title"/>
    </class>

</hibernate-mapping>

Just as with the id element, the name attribute of the property element tells Hibernate which getter and setter methods to use. So, in this case, Hibernate will look for getDate()/setDate(), as well as getTitle()/setTitle().

Why does the date property mapping include the column attribute, but the title doesn't? Without the column attribute Hibernate by default uses the property name as the column name. This works fine for title. However, date is a reserved keyword in most database, so we better map it to a different name.

The next interesting thing is that the title mapping also lacks a type attribute. The types we declare and use in the mapping files are not, as you might expect, Java data types. They are also not SQL database types. These types are so called Hibernate mapping types, converters which can translate from Java to SQL data types and vice versa. Again, Hibernate will try to determine the correct conversion and mapping type itself if the type attribute is not present in the mapping. In some cases this automatic detection (using Reflection on the Java class) might not have the default you expect or need. This is the case with the date property. Hibernate can't know if the property (which is of java.util.Date) should map to a SQL date, timestamp, or time column. We preserve full date and time information by mapping the property with a timestamp converter.

This mapping file should be saved as Event.hbm.xml, right in the directory next to the Event Java class source file. The naming of mapping files can be arbitrary, however the hbm.xml suffix is a convention in the Hibernate developer community. The directory structure should now look like this:

.
+lib
  <Hibernate and third-party libraries>
+src
  +events
    Event.java
    Event.hbm.xml

We continue with the main configuration of Hibernate.

We now have a persistent class and its mapping file in place. It is time to configure Hibernate. Before we do this, we will need a database. HSQL DB, a java-based SQL DBMS, can be downloaded from the HSQL DB website. Actually, you only need the hsqldb.jar from this download. Place this file in the lib/ directory of the development folder.

Create a directory called data in the root of the development directory - this is where HSQL DB will store its data files. Now start the database by running java -classpath lib/hsqldb.jar org.hsqldb.Server in this data directory. You can see it start up and bind to a TCP/IP socket, this is where our application will connect later. If you want to start with a fresh database during this tutorial, shutdown HSQL DB (press CTRL + C in the window), delete all files in the data/ directory, and start HSQL DB again.

Hibernate is the layer in your application which connects to this database, so it needs connection information. The connections are made through a JDBC connection pool, which we also have to configure. The Hibernate distribution contains several open source JDBC connection pooling tools, but will use the Hibernate built-in connection pool for this tutorial. Note that you have to copy the required library into your classpath and use different connection pooling settings if you want to use a production-quality third party JDBC pooling software.

For Hibernate's configuration, we can use a simple hibernate.properties file, a slightly more sophisticated hibernate.cfg.xml file, or even complete programmatic setup. Most users prefer the XML configuration file:

<?xml version='1.0' encoding='utf-8'?>
<!DOCTYPE hibernate-configuration PUBLIC
        "-//Hibernate/Hibernate Configuration DTD 3.0//EN"
        "http://hibernate.sourceforge.net/hibernate-configuration-3.0.dtd">

<hibernate-configuration>

    <session-factory>

        <!-- Database connection settings -->
        <property name="connection.driver_class">org.hsqldb.jdbcDriver</property>
        <property name="connection.url">jdbc:hsqldb:hsql://localhost</property>
        <property name="connection.username">sa</property>
        <property name="connection.password"></property>

        <!-- JDBC connection pool (use the built-in) -->
        <property name="connection.pool_size">1</property>

        <!-- SQL dialect -->
        <property name="dialect">org.hibernate.dialect.HSQLDialect</property>

        <!-- Enable Hibernate's automatic session context management -->
        <property name="current_session_context_class">thread</property>

        <!-- Disable the second-level cache  -->
        <property name="cache.provider_class">org.hibernate.cache.NoCacheProvider</property>

        <!-- Echo all executed SQL to stdout -->
        <property name="show_sql">true</property>

        <!-- Drop and re-create the database schema on startup -->
        <property name="hbm2ddl.auto">create</property>

        <mapping resource="events/Event.hbm.xml"/>

    </session-factory>

</hibernate-configuration>

Note that this XML configuration uses a different DTD. We configure Hibernate's SessionFactory - a global factory responsible for a particular database. If you have several databases, use several <session-factory> configurations, usually in several configuration files (for easier startup).

The first four property elements contain the necessary configuration for the JDBC connection. The dialect property element specifies the particular SQL variant Hibernate generates. Hibernate's automatic session management for persistence contexts will come in handy as you will soon see. The hbm2ddl.auto option turns on automatic generation of database schemas - directly into the database. This can of course also be turned off (by removing the config option) or redirected to a file with the help of the SchemaExport Ant task. Finally, we add the mapping file(s) for persistent classes to the configuration.

Copy this file into the source directory, so it will end up in the root of the classpath. Hibernate automatically looks for a file called hibernate.cfg.xml in the root of the classpath, on startup.

We'll now build the tutorial with Ant. You will need to have Ant installed - get it from the Ant download page. How to install Ant will not be covered here. Please refer to the Ant manual. After you have installed Ant, we can start to create the buildfile. It will be called build.xml and placed directly in the development directory.

A basic build file looks like this:

<project name="hibernate-tutorial" default="compile">

    <property name="sourcedir" value="${basedir}/src"/>
    <property name="targetdir" value="${basedir}/bin"/>
    <property name="librarydir" value="${basedir}/lib"/>

    <path id="libraries">
        <fileset dir="${librarydir}">
            <include name="*.jar"/>
        </fileset>
    </path>

    <target name="clean">
        <delete dir="${targetdir}"/>
        <mkdir dir="${targetdir}"/>
    </target>

    <target name="compile" depends="clean, copy-resources">
      <javac srcdir="${sourcedir}"
             destdir="${targetdir}"
             classpathref="libraries"/>
    </target>

    <target name="copy-resources">
        <copy todir="${targetdir}">
            <fileset dir="${sourcedir}">
                <exclude name="**/*.java"/>
            </fileset>
        </copy>
    </target>

</project>

This will tell Ant to add all files in the lib directory ending with .jar to the classpath used for compilation. It will also copy all non-Java source files to the target directory, e.g. configuration and Hibernate mapping files. If you now run Ant, you should get this output:

C:\hibernateTutorial\>ant
Buildfile: build.xml

copy-resources:
     [copy] Copying 2 files to C:\hibernateTutorial\bin

compile:
    [javac] Compiling 1 source file to C:\hibernateTutorial\bin

BUILD SUCCESSFUL
Total time: 1 second 

It's time to load and store some Event objects, but first we have to complete the setup with some infrastructure code. We have to startup Hibernate. This startup includes building a global SessionFactory object and to store it somewhere for easy access in application code. A SessionFactory can open up new Session's. A Session represents a single-threaded unit of work, the SessionFactory is a thread-safe global object, instantiated once.

We'll create a HibernateUtil helper class which takes care of startup and makes accessing a SessionFactory convenient. Let's have a look at the implementation:

package util;

import org.hibernate.*;
import org.hibernate.cfg.*;

public class HibernateUtil {

    private static final SessionFactory sessionFactory;

    static {
        try {
            // Create the SessionFactory from hibernate.cfg.xml
            sessionFactory = new Configuration().configure().buildSessionFactory();
        } catch (Throwable ex) {
            // Make sure you log the exception, as it might be swallowed
            System.err.println("Initial SessionFactory creation failed." + ex);
            throw new ExceptionInInitializerError(ex);
        }
    }

    public static SessionFactory getSessionFactory() {
        return sessionFactory;
    }

}

This class does not only produce the global SessionFactory in its static initializer (called once by the JVM when the class is loaded), but also hides the fact that it uses a static singleton. It might as well lookup the SessionFactory from JNDI in an application server.

If you give the SessionFactory a name in your configuration file, Hibernate will in fact try to bind it to JNDI after it has been built. To avoid this code completely you could also use JMX deployment and let the JMX-capable container instantiate and bind a HibernateService to JNDI. These advanced options are discussed in the Hibernate reference documentation.

Place HibernateUtil.java in the development source directory, in a package next to events:

.
+lib
  <Hibernate and third-party libraries>
+src
  +events
    Event.java
    Event.hbm.xml
  +util
    HibernateUtil.java
  hibernate.cfg.xml
+data
build.xml

This should again compile without problems. We finally need to configure a logging system - Hibernate uses commons logging and leaves you the choice between Log4j and JDK 1.4 logging. Most developers prefer Log4j: copy log4j.properties from the Hibernate distribution (it's in the etc/ directory) to your src directory, next to hibernate.cfg.xml. Have a look at the example configuration and change the settings if you like to have more verbose output. By default, only Hibernate startup message are shown on stdout.

The tutorial infrastructure is complete - and we are ready to do some real work with Hibernate.

Finally, we can use Hibernate to load and store objects. We write an EventManager class with a main() method:

package events;
import org.hibernate.Session;

import java.util.Date;

import util.HibernateUtil;

public class EventManager {

    public static void main(String[] args) {
        EventManager mgr = new EventManager();

        if (args[0].equals("store")) {
            mgr.createAndStoreEvent("My Event", new Date());
        }

        HibernateUtil.getSessionFactory().close();
    }

    private void createAndStoreEvent(String title, Date theDate) {

        Session session = HibernateUtil.getSessionFactory().getCurrentSession();

        session.beginTransaction();

        Event theEvent = new Event();
        theEvent.setTitle(title);
        theEvent.setDate(theDate);

        session.save(theEvent);

        session.getTransaction().commit();
    }

}

We create a new Event object, and hand it over to Hibernate. Hibernate now takes care of the SQL and executes INSERTs on the database. Let's have a look at the Session and Transaction-handling code before we run this.

A Session is a single unit of work. For now we'll keep things simple and assume a one-to-one granularity between a Hibernate Session and a database transaction. To shield our code from the actual underlying transaction system (in this case plain JDBC, but it could also run with JTA) we use the Transaction API that is available on the Hibernate Session.

What does sessionFactory.getCurrentSession() do? First, you can call it as many times and anywhere you like, once you get hold of your SessionFactory (easy thanks to HibernateUtil). The getCurrentSession() method always returns the "current" unit of work. Remember that we switched the configuration option for this mechanism to "thread" in hibernate.cfg.xml? Hence, the scope of the current unit of work is the current Java thread that executes our application. However, this is not the full truth. A Session begins when it is first needed, when the first call to getCurrentSession() is made. It is then bound by Hibernate to the current thread. When the transaction ends, either committed or rolled back, Hibernate also unbinds the Session from the thread and closes it for you. If you call getCurrentSession() again, you get a new Session and can start a new unit of work. This thread-bound programming model is the most popular way of using Hibernate.

Have a look at Chapter 11, Transactions And Concurrency for more information about transaction handling and demarcation. We also skipped any error handling and rollback in the previous example.

To run this first routine we have to add a callable target to the Ant build file:

<target name="run" depends="compile">
    <java fork="true" classname="events.EventManager" classpathref="libraries">
        <classpath path="${targetdir}"/>
        <arg value="${action}"/>
    </java>
</target>

The value of the action argument is set on the command line when calling the target:

C:\hibernateTutorial\>ant run -Daction=store

You should see, after compilation, Hibernate starting up and, depending on your configuration, lots of log output. At the end you will find the following line:

[java] Hibernate: insert into EVENTS (EVENT_DATE, title, EVENT_ID) values (?, ?, ?)

This is the INSERT executed by Hibernate, the question marks represent JDBC bind parameters. To see the values bound as arguments, or to reduce the verbosity of the log, check your log4j.properties.

Now we'd like to list stored events as well, so we add an option to the main method:

if (args[0].equals("store")) {
    mgr.createAndStoreEvent("My Event", new Date());
}
else if (args[0].equals("list")) {
    List events = mgr.listEvents();
    for (int i = 0; i < events.size(); i++) {
        Event theEvent = (Event) events.get(i);
        System.out.println("Event: " + theEvent.getTitle() +
                           " Time: " + theEvent.getDate());
    }
}

We also add a new listEvents() method:

private List listEvents() {

    Session session = HibernateUtil.getSessionFactory().getCurrentSession();

    session.beginTransaction();

    List result = session.createQuery("from Event").list();

    session.getTransaction().commit();

    return result;
}

What we do here is use an HQL (Hibernate Query Language) query to load all existing Event objects from the database. Hibernate will generate the appropriate SQL, send it to the database and populate Event objects with the data. You can create more complex queries with HQL, of course.

Now, to execute and test all of this, follow these steps:

If you now call Ant with -Daction=list, you should see the events you have stored so far. You can of course also call the store action a few times more.

Note: Most new Hibernate users fail at this point and we see questions about Table not found error messages regularly. However, if you follow the steps outlined above you will not have this problem, as hbm2ddl creates the database schema on the first run, and subsequent application restarts will use this schema. If you change the mapping and/or database schema, you have to re-enable hbm2ddl once again.

The first cut of the Person class is simple:

package events;

public class Person {

    private Long id;
    private int age;
    private String firstname;
    private String lastname;

    public Person() {}

    // Accessor methods for all properties, private setter for 'id'

}

Create a new mapping file called Person.hbm.xml (don't forget the DTD reference at the top):

<hibernate-mapping>

    <class name="events.Person" table="PERSON">
        <id name="id" column="PERSON_ID">
            <generator class="native"/>
        </id>
        <property name="age"/>
        <property name="firstname"/>
        <property name="lastname"/>
    </class>

</hibernate-mapping>

Finally, add the new mapping to Hibernate's configuration:

<mapping resource="events/Event.hbm.xml"/>
<mapping resource="events/Person.hbm.xml"/>

We'll now create an association between these two entities. Obviously, persons can participate in events, and events have participants. The design questions we have to deal with are: directionality, multiplicity, and collection behavior.

We'll add a collection of events to the Person class. That way we can easily navigate to the events for a particular person, without executing an explicit query - by calling aPerson.getEvents(). We use a Java collection, a Set, because the collection will not contain duplicate elements and the ordering is not relevant for us.

We need a unidirectional, many-valued associations, implemented with a Set. Let's write the code for this in the Java classes and then map it:

public class Person {

    private Set events = new HashSet();

    public Set getEvents() {
        return events;
    }

    public void setEvents(Set events) {
        this.events = events;
    }
}

Before we map this association, think about the other side. Clearly, we could just keep this unidirectional. Or, we could create another collection on the Event, if we want to be able to navigate it bi-directional, i.e. anEvent.getParticipants(). This is not necessary, from a functional perspective. You could always execute an explicit query to retrieve the participants for a particular event. This is a design choice left to you, but what is clear from this discussion is the multiplicity of the association: "many" valued on both sides, we call this a many-to-many association. Hence, we use Hibernate's many-to-many mapping:

<class name="events.Person" table="PERSON">
    <id name="id" column="PERSON_ID">
        <generator class="native"/>
    </id>
    <property name="age"/>
    <property name="firstname"/>
    <property name="lastname"/>

    <set name="events" table="PERSON_EVENT">
        <key column="PERSON_ID"/>
        <many-to-many column="EVENT_ID" class="events.Event"/>
    </set>

</class>

Hibernate supports all kinds of collection mappings, a <set> being most common. For a many-to-many association (or n:m entity relationship), an association table is needed. Each row in this table represents a link between a person and an event. The table name is configured with the table attribute of the set element. The identifier column name in the association, for the person's side, is defined with the <key> element, the column name for the event's side with the column attribute of the <many-to-many>. You also have to tell Hibernate the class of the objects in your collection (correct: the class on the other side of the collection of references).

The database schema for this mapping is therefore:

    _____________        __________________
   |             |      |                  |       _____________
   |   EVENTS    |      |   PERSON_EVENT   |      |             |
   |_____________|      |__________________|      |    PERSON   |
   |             |      |                  |      |_____________|
   | *EVENT_ID   | <--> | *EVENT_ID        |      |             |
   |  EVENT_DATE |      | *PERSON_ID       | <--> | *PERSON_ID  |
   |  TITLE      |      |__________________|      |  AGE        |
   |_____________|                                |  FIRSTNAME  |
                                                  |  LASTNAME   |
                                                  |_____________|
 

Let's bring some people and events together in a new method in EventManager:

private void addPersonToEvent(Long personId, Long eventId) {

    Session session = HibernateUtil.getSessionFactory().getCurrentSession();
    session.beginTransaction();

    Person aPerson = (Person) session.load(Person.class, personId);
    Event anEvent = (Event) session.load(Event.class, eventId);

    aPerson.getEvents().add(anEvent);

    session.getTransaction().commit();
}

After loading a Person and an Event, simply modify the collection using the normal collection methods. As you can see, there is no explicit call to update() or save(), Hibernate automatically detects that the collection has been modified and needs to be updated. This is called automatic dirty checking, and you can also try it by modifying the name or the date property of any of your objects. As long as they are in persistent state, that is, bound to a particular Hibernate Session (i.e. they have been just loaded or saved in a unit of work), Hibernate monitors any changes and executes SQL in a write-behind fashion. The process of synchronizing the memory state with the database, usually only at the end of a unit of work, is called flushing. In our code, the unit of work ends with a commit (or rollback) of the database transaction - as defined by the thread configuration option for the CurrentSessionContext class.

You might of course load person and event in different units of work. Or you modify an object outside of a Session, when it is not in persistent state (if it was persistent before, we call this state detached). You can even modify a collection when it is detached:

private void addPersonToEvent(Long personId, Long eventId) {

    Session session = HibernateUtil.getSessionFactory().getCurrentSession();
    session.beginTransaction();

    Person aPerson = (Person) session
            .createQuery("select p from Person p left join fetch p.events where p.id = :pid")
            .setParameter("pid", personId)
            .uniqueResult(); // Eager fetch the collection so we can use it detached

    Event anEvent = (Event) session.load(Event.class, eventId);

    session.getTransaction().commit();

    // End of first unit of work

    aPerson.getEvents().add(anEvent); // aPerson (and its collection) is detached

    // Begin second unit of work

    Session session2 = HibernateUtil.getSessionFactory().getCurrentSession();
    session2.beginTransaction();

    session2.update(aPerson); // Reattachment of aPerson

    session2.getTransaction().commit();
}

The call to update makes a detached object persistent again, you could say it binds it to a new unit of work, so any modifications you made to it while detached can be saved to the database. This includes any modifications (additions/deletions) you made to a collection of that entity object.

Well, this is not much use in our current situation, but it's an important concept you can design into your own application. For now, complete this exercise by adding a new action to the EventManager's main method and call it from the command line. If you need the identifiers of a person and an event - the save() method returns it (you might have to modify some of the previous methods to return that identifier):

else if (args[0].equals("addpersontoevent")) {
    Long eventId = mgr.createAndStoreEvent("My Event", new Date());
    Long personId = mgr.createAndStorePerson("Foo", "Bar");
    mgr.addPersonToEvent(personId, eventId);
    System.out.println("Added person " + personId + " to event " + eventId);

This was an example of an association between two equally important classes, two entities. As mentioned earlier, there are other classes and types in a typical model, usually "less important". Some you have already seen, like an int or a String. We call these classes value types, and their instances depend on a particular entity. Instances of these types don't have their own identity, nor are they shared between entities (two persons don't reference the same firstname object, even if they have the same first name). Of course, value types can not only be found in the JDK (in fact, in a Hibernate application all JDK classes are considered value types), but you can also write dependent classes yourself, Address or MonetaryAmount, for example.

You can also design a collection of value types. This is conceptually very different from a collection of references to other entities, but looks almost the same in Java.

We add a collection of value typed objects to the Person entity. We want to store email addresses, so the type we use is String, and the collection is again a Set:

private Set emailAddresses = new HashSet();

public Set getEmailAddresses() {
    return emailAddresses;
}

public void setEmailAddresses(Set emailAddresses) {
    this.emailAddresses = emailAddresses;
}

The mapping of this Set:

<set name="emailAddresses" table="PERSON_EMAIL_ADDR">
    <key column="PERSON_ID"/>
    <element type="string" column="EMAIL_ADDR"/>
</set>

The difference compared with the earlier mapping is the element part, which tells Hibernate that the collection does not contain references to another entity, but a collection of elements of type String (the lowercase name tells you it's a Hibernate mapping type/converter). Once again, the table attribute of the set element determines the table name for the collection. The key element defines the foreign-key column name in the collection table. The column attribute in the element element defines the column name where the String values will actually be stored.

Have a look at the updated schema:

  _____________        __________________
 |             |      |                  |       _____________
 |   EVENTS    |      |   PERSON_EVENT   |      |             |       ___________________
 |_____________|      |__________________|      |    PERSON   |      |                   |
 |             |      |                  |      |_____________|      | PERSON_EMAIL_ADDR |
 | *EVENT_ID   | <--> | *EVENT_ID        |      |             |      |___________________|
 |  EVENT_DATE |      | *PERSON_ID       | <--> | *PERSON_ID  | <--> |  *PERSON_ID       |
 |  TITLE      |      |__________________|      |  AGE        |      |  *EMAIL_ADDR      |
 |_____________|                                |  FIRSTNAME  |      |___________________|
                                                |  LASTNAME   |
                                                |_____________|
 

You can see that the primary key of the collection table is in fact a composite key, using both columns. This also implies that there can't be duplicate email addresses per person, which is exactly the semantics we need for a set in Java.

You can now try and add elements to this collection, just like we did before by linking persons and events. It's the same code in Java:

private void addEmailToPerson(Long personId, String emailAddress) {

    Session session = HibernateUtil.getSessionFactory().getCurrentSession();
    session.beginTransaction();

    Person aPerson = (Person) session.load(Person.class, personId);

    // The getEmailAddresses() might trigger a lazy load of the collection
    aPerson.getEmailAddresses().add(emailAddress);

    session.getTransaction().commit();
}

This time we didnt' use a fetch query to initialize the collection. Hence, the call to its getter method will trigger an additional select to initialize it, so we can add an element to it. Monitor the SQL log and try to optimize this with an eager fetch.

Next we are going to map a bi-directional association - making the association between person and event work from both sides in Java. Of course, the database schema doesn't change, we still have many-to-many multiplicity. A relational database is more flexible than a network programming language, so it doesn't need anything like a navigation direction - data can be viewed and retrieved in any possible way.

First, add a collection of participants to the Event Event class:

private Set participants = new HashSet();

public Set getParticipants() {
    return participants;
}

public void setParticipants(Set participants) {
    this.participants = participants;
}

Now map this side of the association too, in Event.hbm.xml.

<set name="participants" table="PERSON_EVENT" inverse="true">
    <key column="EVENT_ID"/>
    <many-to-many column="PERSON_ID" class="events.Person"/>
</set>

As you see, these are normal set mappings in both mapping documents. Notice that the column names in key and many-to-many are swapped in both mapping documents. The most important addition here is the inverse="true" attribute in the set element of the Event's collection mapping.

What this means is that Hibernate should take the other side - the Person class - when it needs to find out information about the link between the two. This will be a lot easier to understand once you see how the bi-directional link between our two entities is created .

First, keep in mind that Hibernate does not affect normal Java semantics. How did we create a link between a Person and an Event in the unidirectional example? We added an instance of Event to the collection of event references, of an instance of Person. So, obviously, if we want to make this link working bi-directional, we have to do the same on the other side - adding a Person reference to the collection in an Event. This "setting the link on both sides" is absolutely necessary and you should never forget doing it.

Many developers program defensive and create a link management methods to correctly set both sides, e.g. in Person:

protected Set getEvents() {
    return events;
}

protected void setEvents(Set events) {
    this.events = events;
}

public void addToEvent(Event event) {
    this.getEvents().add(event);
    event.getParticipants().add(this);
}

public void removeFromEvent(Event event) {
    this.getEvents().remove(event);
    event.getParticipants().remove(this);
}

Notice that the get and set methods for the collection are now protected - this allows classes in the same package and subclasses to still access the methods, but prevents everybody else from messing with the collections directly (well, almost). You should probably do the same with the collection on the other side.

What about the inverse mapping attribute? For you, and for Java, a bi-directional link is simply a matter of setting the references on both sides correctly. Hibernate however doesn't have enough information to correctly arrange SQL INSERT and UPDATE statements (to avoid constraint violations), and needs some help to handle bi-directional associations properly. Making one side of the association inverse tells Hibernate to basically ignore it, to consider it a mirror of the other side. That's all that is necessary for Hibernate to work out all of the issues when transformation a directional navigation model to a SQL database schema. The rules you have to remember are straightforward: All bi-directional associations need one side as inverse. In a one-to-many association it has to be the many-side, in many-to-many association you can pick either side, there is no difference.

Create a new class in your source directory, in the events package:

package events;

// Imports

public class EventManagerServlet extends HttpServlet {

    private final SimpleDateFormat dateFormatter =
                            new SimpleDateFormat("dd.MM.yyyy");

    // Servlet code
}

The dateFormatter is a tool we'll need later to convert Date objects from and to strings. It makes sense to only have one formatter as a member of the servlet.

The servlet handles HTTP GET requests only, hence, the method we implement is doGet():

protected void doGet(HttpServletRequest request,
                     HttpServletResponse response)
        throws ServletException, IOException {

    try {
        // Begin unit of work
        HibernateUtil.getSessionFactory()
                .getCurrentSession().beginTransaction();

        // Process request and render page...

        // End unit of work
        HibernateUtil.getSessionFactory()
                .getCurrentSession().getTransaction().commit();

    } catch (Exception ex) {
        HibernateUtil.getSessionFactory()
                .getCurrentSession().getTransaction().rollback();
        throw new ServletException(ex);
    }

}

The pattern we are applying here is called session-per-request. When a request hits the servlet, a new Hibernate Session is opened through the first call to getCurrentSession() on the SessionFactory. Then a database transaction is started—all data access as to occur inside a transaction, no matter if data is read or written (we don't use the auto-commit mode in applications).

Next, the possible actions of the request are processed and the response HTML is rendered. We'll get to that part soon.

Finally, the unit of work ends when processing and rendering is complete. If any problem occured during processing or rendering, an exception will be thrown and the database transaction rolled back. This completes the session-per-request pattern. Instead of the transaction demarcation code in every servlet you could also write a servlet filter. See the Hibernate website and Wiki for more information about this pattern, called Open Session in View—you'll need it as soon as you consider rendering your view in JSP, not in a servlet.

Let's implement the processing of the request and rendering of the page.

// Write HTML header
PrintWriter out = response.getWriter();
out.println("<html><head><title>Event Manager</title></head><body>");

// Handle actions
if ( "store".equals(request.getParameter("action")) ) {

    String eventTitle = request.getParameter("eventTitle");
    String eventDate = request.getParameter("eventDate");

    if ( "".equals(eventTitle) || "".equals(eventDate) ) {
        out.println("<b><i>Please enter event title and date.</i></b>");
    } else {
        createAndStoreEvent(eventTitle, dateFormatter.parse(eventDate));
        out.println("<b><i>Added event.</i></b>");
    }
}

// Print page
printEventForm(out);
listEvents(out);

// Write HTML footer
out.println("</body></html>");
out.flush();
out.close();

Granted, this coding style with a mix of Java and HTML would not scale in a more complex application—keep in mind that we are only illustrating basic Hibernate concepts in this tutorial. The code prints an HTML header and a footer. Inside this page, an HTML form for event entry and a list of all events in the database are printed. The first method is trivial and only outputs HTML:

private void printEventForm(PrintWriter out) {
    out.println("<h2>Add new event:</h2>");
    out.println("<form>");
    out.println("Title: <input name='eventTitle' length='50'/><br/>");
    out.println("Date (e.g. 24.12.2009): <input name='eventDate' length='10'/><br/>");
    out.println("<input type='submit' name='action' value='store'/>");
    out.println("</form>");
}

The listEvents() method uses the Hibernate Session bound to the current thread to execute a query:

private void listEvents(PrintWriter out) {
    List result = HibernateUtil.getSessionFactory()
                    .getCurrentSession().createCriteria(Event.class).list();
    if (result.size() > 0) {
        out.println("<h2>Events in database:</h2>");
        out.println("<table border='1'>");
        out.println("<tr>");
        out.println("<th>Event title</th>");
        out.println("<th>Event date</th>");
        out.println("</tr>");
        for (Iterator it = result.iterator(); it.hasNext();) {
            Event event = (Event) it.next();
            out.println("<tr>");
            out.println("<td>" + event.getTitle() + "</td>");
            out.println("<td>" + dateFormatter.format(event.getDate()) + "</td>");
            out.println("</tr>");
        }
        out.println("</table>");
    }
}

Finally, the store action is dispatched to the createAndStoreEvent() method, which also uses the Session of the current thread:

protected void createAndStoreEvent(String title, Date theDate) {
    Event theEvent = new Event();
    theEvent.setTitle(title);
    theEvent.setDate(theDate);

    HibernateUtil.getSessionFactory()
                    .getCurrentSession().save(theEvent);
}

That's it, the servlet is complete. A request to the servlet will be processed in a single Session and Transaction. As earlier in the standalone application, Hibernate can automatically bind these ojects to the current thread of execution. This gives you the freedom to layer your code and access the SessionFactory in any way you like. Usually you'd use a more sophisticated design and move the data access code into data access objects (the DAO pattern). See the Hibernate Wiki for more examples.

To deploy this application you have to create a web archive, a WAR. Add the following Ant target to your build.xml:

<target name="war" depends="compile">
    <war destfile="hibernate-tutorial.war" webxml="web.xml">
        <lib dir="${librarydir}">
          <exclude name="jsdk*.jar"/>
        </lib>

        <classes dir="${targetdir}"/>
    </war>
</target>

This target creates a file called hibernate-tutorial.war in your project directory. It packages all libraries and the web.xml descriptor, which is expected in the base directory of your project:

<?xml version="1.0" encoding="UTF-8"?>
<web-app version="2.4"
    xmlns="http://java.sun.com/xml/ns/j2ee"
    xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance"
    xsi:schemaLocation="http://java.sun.com/xml/ns/j2ee http://java.sun.com/xml/ns/j2ee/web-app_2_4.xsd">

    <servlet>
        <servlet-name>Event Manager</servlet-name>
        <servlet-class>events.EventManagerServlet</servlet-class>
    </servlet>

    <servlet-mapping>
        <servlet-name>Event Manager</servlet-name>
        <url-pattern>/eventmanager</url-pattern>
    </servlet-mapping>
</web-app>

Before you compile and deploy the web application, note that an additional library is required: jsdk.jar. This is the Java servlet development kit, if you don't have this library already, get it from the Sun website and copy it to your library directory. However, it will be only used for compliation and excluded from the WAR package.

To build and deploy call ant war in your project directory and copy the hibernate-tutorial.war file into your Tomcat webapp directory. If you don't have Tomcat installed, download it and follow the installation instructions. You don't have to change any Tomcat configuration to deploy this application though.

Once deployed and Tomcat is running, access the application at http://localhost:8080/hibernate-tutorial/eventmanager. Make sure you watch the Tomcat log to see Hibernate initialize when the first request hits your servlet (the static initializer in HibernateUtil is called) and to get the detailed output if any exceptions occurs.

You should always set the hibernate.dialect property to the correct org.hibernate.dialect.Dialect subclass for your database. If you specify a dialect, Hibernate will use sensible defaults for some of the other properties listed above, saving you the effort of specifying them manually.

If your database supports ANSI, Oracle or Sybase style outer joins, outer join fetching will often increase performance by limiting the number of round trips to and from the database (at the cost of possibly more work performed by the database itself). Outer join fetching allows a whole graph of objects connected by many-to-one, one-to-many, many-to-many and one-to-one associations to be retrieved in a single SQL SELECT.

Outer join fetching may be disabled globally by setting the property hibernate.max_fetch_depth to 0. A setting of 1 or higher enables outer join fetching for one-to-one and many-to-one associations which have been mapped with fetch="join".

See Section 19.1, “Fetching strategies” for more information.

Oracle limits the size of byte arrays that may be passed to/from its JDBC driver. If you wish to use large instances of binary or serializable type, you should enable hibernate.jdbc.use_streams_for_binary. This is a system-level setting only.

The properties prefixed by hibernate.cache allow you to use a process or cluster scoped second-level cache system with Hibernate. See the Section 19.2, “The Second Level Cache” for more details.

You may define new Hibernate query tokens using hibernate.query.substitutions. For example:

hibernate.query.substitutions true=1, false=0

would cause the tokens true and false to be translated to integer literals in the generated SQL.

hibernate.query.substitutions toLowercase=LOWER

would allow you to rename the SQL LOWER function.

If you enable hibernate.generate_statistics, Hibernate will expose a number of metrics that are useful when tuning a running system via SessionFactory.getStatistics(). Hibernate can even be configured to expose these statistics via JMX. Read the Javadoc of the interfaces in org.hibernate.stats for more information.

The Hibernate Session API is independent of any transaction demarcation system in your architecture. If you let Hibernate use JDBC directly, through a connection pool, you may begin and end your transactions by calling the JDBC API. If you run in a J2EE application server, you might want to use bean-managed transactions and call the JTA API and UserTransaction when needed.

To keep your code portable between these two (and other) environments we recommend the optional Hibernate Transaction API, which wraps and hides the underlying system. You have to specify a factory class for Transaction instances by setting the Hibernate configuration property hibernate.transaction.factory_class.

There are three standard (built-in) choices:

You may also define your own transaction strategies (for a CORBA transaction service, for example).

Some features in Hibernate (i.e. the second level cache, Contextual Sessions with JTA, etc.) require access to the JTA TransactionManager in a managed environment. In an application server you have to specify how Hibernate should obtain a reference to the TransactionManager, since J2EE does not standardize a single mechanism:

A JNDI bound Hibernate SessionFactory can simplify the lookup of the factory and the creation of new Sessions. Note that this is not related to a JNDI bound Datasource, both simply use the same registry!

If you wish to have the SessionFactory bound to a JNDI namespace, specify a name (eg. java:hibernate/SessionFactory) using the property hibernate.session_factory_name. If this property is omitted, the SessionFactory will not be bound to JNDI. (This is especially useful in environments with a read-only JNDI default implementation, e.g. Tomcat.)

When binding the SessionFactory to JNDI, Hibernate will use the values of hibernate.jndi.url, hibernate.jndi.class to instantiate an initial context. If they are not specified, the default InitialContext will be used.

Hibernate will automatically place the SessionFactory in JNDI after you call cfg.buildSessionFactory(). This means you will at least have this call in some startup code (or utility class) in your application, unless you use JMX deployment with the HibernateService (discussed later).

If you use a JNDI SessionFactory, an EJB or any other class may obtain the SessionFactory using a JNDI lookup.

We recommend that you bind the SessionFactory to JNDI in a managend environment and use a static singleton otherwise. To shield your application code from these details, we also recommend to hide the actual lookup code for a SessionFactory in a helper class, such as HibernateUtil.getSessionFactory(). Note that such a class is also a convenient way to startup Hibernate—see chapter 1.

The easiest way to handle Sessions and transactions is Hibernates automatic "current" Session management. See the discussion of Section 2.5, “Contextual Sessions”. Using the "jta" session context, if there is no Hibernate Session associated with the current JTA transaction, one will be started and associated with that JTA transaction the first time you call sessionFactory.getCurrentSession(). The Sessions retrieved via getCurrentSession() in "jta" context will be set to automatically flush before the transaction completes, close after the transaction completes, and aggressively release JDBC connections after each statement. This allows the Sessions to be managed by the lifecycle of the JTA transaction to which it is associated, keeping user code clean of such management concerns. Your code can either use JTA programmatically through UserTransaction, or (recommended for portable code) use the Hibernate Transaction API to set transaction boundaries. If you run in an EJB container, declarative transaction demarcation with CMT is preferred.

The line cfg.buildSessionFactory() still has to be executed somewhere to get a SessionFactory into JNDI. You can do this either in a static initializer block (like the one in HibernateUtil) or you deploy Hibernate as a managed service.

Hibernate is distributed with org.hibernate.jmx.HibernateService for deployment on an application server with JMX capabilities, such as JBoss AS. The actual deployment and configuration is vendor specific. Here is an example jboss-service.xml for JBoss 4.0.x:

<?xml version="1.0"?>
<server>

<mbean code="org.hibernate.jmx.HibernateService"
    name="jboss.jca:service=HibernateFactory,name=HibernateFactory">

    <!-- Required services -->
    <depends>jboss.jca:service=RARDeployer</depends>
    <depends>jboss.jca:service=LocalTxCM,name=HsqlDS</depends>

    <!-- Bind the Hibernate service to JNDI -->
    <attribute name="JndiName">java:/hibernate/SessionFactory</attribute>

    <!-- Datasource settings -->
    <attribute name="Datasource">java:HsqlDS</attribute>
    <attribute name="Dialect">org.hibernate.dialect.HSQLDialect</attribute>

    <!-- Transaction integration -->
    <attribute name="TransactionStrategy">
        org.hibernate.transaction.JTATransactionFactory</attribute>
    <attribute name="TransactionManagerLookupStrategy">
        org.hibernate.transaction.JBossTransactionManagerLookup</attribute>
    <attribute name="FlushBeforeCompletionEnabled">true</attribute>
    <attribute name="AutoCloseSessionEnabled">true</attribute>

    <!-- Fetching options -->
    <attribute name="MaximumFetchDepth">5</attribute>

    <!-- Second-level caching -->
    <attribute name="SecondLevelCacheEnabled">true</attribute>
    <attribute name="CacheProviderClass">org.hibernate.cache.EhCacheProvider</attribute>
    <attribute name="QueryCacheEnabled">true</attribute>

    <!-- Logging -->
    <attribute name="ShowSqlEnabled">true</attribute>

    <!-- Mapping files -->
    <attribute name="MapResources">auction/Item.hbm.xml,auction/Category.hbm.xml</attribute>

</mbean>

</server>

This file is deployed in a directory called META-INF and packaged in a JAR file with the extension .sar (service archive). You also need to package Hibernate, its required third-party libraries, your compiled persistent classes, as well as your mapping files in the same archive. Your enterprise beans (usually session beans) may be kept in their own JAR file, but you may include this EJB JAR file in the main service archive to get a single (hot-)deployable unit. Consult the JBoss AS documentation for more information about JMX service and EJB deployment.

Cat has a no-argument constructor. All persistent classes must have a default constructor (which may be non-public) so that Hibernate can instantiate them using Constructor.newInstance(). We strongly recommend having a default constructor with at least package visibility for runtime proxy generation in Hibernate.

Cat has a property called id. This property maps to the primary key column of a database table. The property might have been called anything, and its type might have been any primitive type, any primitive "wrapper" type, java.lang.String or java.util.Date. (If your legacy database table has composite keys, you can even use a user-defined class with properties of these types - see the section on composite identifiers later.)

The identifier property is strictly optional. You can leave them off and let Hibernate keep track of object identifiers internally. We do not recommend this, however.

In fact, some functionality is available only to classes which declare an identifier property:

We recommend you declare consistently-named identifier properties on persistent classes. We further recommend that you use a nullable (ie. non-primitive) type.

A central feature of Hibernate, proxies, depends upon the persistent class being either non-final, or the implementation of an interface that declares all public methods.

You can persist final classes that do not implement an interface with Hibernate, but you won't be able to use proxies for lazy association fetching - which will limit your options for performance tuning.

You should also avoid declaring public final methods on the non-final classes. If you want to use a class with a public final method, you must explicitly disable proying by setting lazy="false".

Cat declares accessor methods for all its persistent fields. Many other ORM tools directly persist instance variables. We believe it is better to provide an indirection between the relational schema and internal data structures of the class. By default, Hibernate persists JavaBeans style properties, and recognizes method names of the form getFoo, isFoo and setFoo. You may switch to direct field access for particular properties, if needed.

Properties need not be declared public - Hibernate can persist a property with a default, protected or private get / set pair.

All XML mappings should declare the doctype shown. The actual DTD may be found at the URL above, in the directory hibernate-x.x.x/src/org/hibernate or in hibernate3.jar. Hibernate will always look for the DTD in its classpath first. If you experience lookups of the DTD using an Internet connection, check your DTD declaration against the contents of your claspath.

This element has several optional attributes. The schema and catalog attributes specify that tables referred to in this mapping belong to the named schema and/or catalog. If specified, tablenames will be qualified by the given schema and catalog names. If missing, tablenames will be unqualified. The default-cascade attribute specifies what cascade style should be assumed for properties and collections which do not specify a cascade attribute. The auto-import attribute lets us use unqualified class names in the query language, by default.

<hibernate-mapping
         schema="schemaName"                          (1)
         catalog="catalogName"                        (2)
         default-cascade="cascade_style"              (3)
         default-access="field|property|ClassName"    (4)
         default-lazy="true|false"                    (5)
         auto-import="true|false"                     (6)
         package="package.name"                       (7)
 />

If you have two persistent classes with the same (unqualified) name, you should set auto-import="false". Hibernate will throw an exception if you attempt to assign two classes to the same "imported" name.

Note that the hibernate-mapping element allows you to nest several persistent <class> mappings, as shown above. It is however good practice (and expected by some tools) to map only a single persistent class (or a single class hierarchy) in one mapping file and name it after the persistent superclass, e.g. Cat.hbm.xml, Dog.hbm.xml, or if using inheritance, Animal.hbm.xml.

You may declare a persistent class using the class element:

<class
        name="ClassName"                              (1)
        table="tableName"                             (2)
        discriminator-value="discriminator_value"     (3)
        mutable="true|false"                          (4)
        schema="owner"                                (5)
        catalog="catalog"                             (6)
        proxy="ProxyInterface"                        (7)
        dynamic-update="true|false"                   (8)
        dynamic-insert="true|false"                   (9)
        select-before-update="true|false"             (10)
        polymorphism="implicit|explicit"              (11)
        where="arbitrary sql where condition"         (12)
        persister="PersisterClass"                    (13)
        batch-size="N"                                (14)
        optimistic-lock="none|version|dirty|all"      (15)
        lazy="true|false"                             (16)
        entity-name="EntityName"                      (17)
        check="arbitrary sql check condition"         (18)
        rowid="rowid"                                 (19)
        subselect="SQL expression"                    (20)
        abstract="true|false"                         (21)
        node="element-name"
/>

It is perfectly acceptable for the named persistent class to be an interface. You would then declare implementing classes of that interface using the <subclass> element. You may persist any static inner class. You should specify the class name using the standard form ie. eg.Foo$Bar.

Immutable classes, mutable="false", may not be updated or deleted by the application. This allows Hibernate to make some minor performance optimizations.

The optional proxy attribute enables lazy initialization of persistent instances of the class. Hibernate will initially return CGLIB proxies which implement the named interface. The actual persistent object will be loaded when a method of the proxy is invoked. See "Proxies for Lazy Initialization" below.

Implicit polymorphism means that instances of the class will be returned by a query that names any superclass or implemented interface or the class and that instances of any subclass of the class will be returned by a query that names the class itself. Explicit polymorphism means that class instances will be returned only by queries that explicitly name that class and that queries that name the class will return only instances of subclasses mapped inside this <class> declaration as a <subclass> or <joined-subclass>. For most purposes the default, polymorphism="implicit", is appropriate. Explicit polymorphism is useful when two different classes are mapped to the same table (this allows a "lightweight" class that contains a subset of the table columns).

The persister attribute lets you customize the persistence strategy used for the class. You may, for example, specify your own subclass of org.hibernate.persister.EntityPersister or you might even provide a completely new implementation of the interface org.hibernate.persister.ClassPersister that implements persistence via, for example, stored procedure calls, serialization to flat files or LDAP. See org.hibernate.test.CustomPersister for a simple example (of "persistence" to a Hashtable).

Note that the dynamic-update and dynamic-insert settings are not inherited by subclasses and so may also be specified on the <subclass> or <joined-subclass> elements. These settings may increase performance in some cases, but might actually decrease performance in others. Use judiciously.

Use of select-before-update will usually decrease performance. It is very useful to prevent a database update trigger being called unnecessarily if you reattach a graph of detached instances to a Session.

If you enable dynamic-update, you will have a choice of optimistic locking strategies:

We very strongly recommend that you use version/timestamp columns for optimistic locking with Hibernate. This is the optimal strategy with respect to performance and is the only strategy that correctly handles modifications made to detached instances (ie. when Session.merge() is used).

There is no difference between a view and a base table for a Hibernate mapping, as expected this is transparent at the database level (note that some DBMS don't support views properly, especially with updates). Sometimes you want to use a view, but can't create one in the database (ie. with a legacy schema). In this case, you can map an immutable and read-only entity to a given SQL subselect expression:

<class name="Summary">
    <subselect>
        select item.name, max(bid.amount), count(*)
        from item
        join bid on bid.item_id = item.id
        group by item.name
    </subselect>
    <synchronize table="item"/>
    <synchronize table="bid"/>
    <id name="name"/>
    ...
</class>

Declare the tables to synchronize this entity with, ensuring that auto-flush happens correctly, and that queries against the derived entity do not return stale data. The <subselect> is available as both as an attribute and a nested mapping element.

Mapped classes must declare the primary key column of the database table. Most classes will also have a JavaBeans-style property holding the unique identifier of an instance. The <id> element defines the mapping from that property to the primary key column.

<id
        name="propertyName"                                          (1)
        type="typename"                                              (2)
        column="column_name"                                         (3)
        unsaved-value="null|any|none|undefined|id_value"             (4)
        access="field|property|ClassName">                           (5)
        node="element-name|@attribute-name|element/@attribute|."

        <generator class="generatorClass"/>
</id>

If the name attribute is missing, it is assumed that the class has no identifier property.

The unsaved-value attribute is almost never needed in Hibernate3.

There is an alternative <composite-id> declaration to allow access to legacy data with composite keys. We strongly discourage its use for anything else.

<id name="id" type="long" column="cat_id">
        <generator class="org.hibernate.id.TableHiLoGenerator">
                <param name="table">uid_table</param>
                <param name="column">next_hi_value_column</param>
        </generator>
</id>

<id name="id" type="long" column="cat_id">
        <generator class="hilo">
                <param name="table">hi_value</param>
                <param name="column">next_value</param>
                <param name="max_lo">100</param>
        </generator>
</id>

<id name="id" type="long" column="cat_id">
        <generator class="seqhilo">
                <param name="sequence">hi_value</param>
                <param name="max_lo">100</param>
        </generator>
</id>

<id name="id" type="long" column="person_id">
        <generator class="sequence">
                <param name="sequence">person_id_sequence</param>
        </generator>
</id>

<id name="id" type="long" column="person_id" unsaved-value="0">
        <generator class="identity"/>
</id>

<id name="id" type="long" column="person_id">
        <generator class="select">
                <param name="key">socialSecurityNumber</param>
        </generator>
</id>

<composite-id
        name="propertyName"
        class="ClassName"
        mapped="true|false"
        access="field|property|ClassName">
        node="element-name|."

        <key-property name="propertyName" type="typename" column="column_name"/>
        <key-many-to-one name="propertyName class="ClassName" column="column_name"/>
        ......
</composite-id>

For a table with a composite key, you may map multiple properties of the class as identifier properties. The <composite-id> element accepts <key-property> property mappings and <key-many-to-one> mappings as child elements.

<composite-id>
        <key-property name="medicareNumber"/>
        <key-property name="dependent"/>
</composite-id>

Your persistent class must override equals() and hashCode() to implement composite identifier equality. It must also implements Serializable.

Unfortunately, this approach to composite identifiers means that a persistent object is its own identifier. There is no convenient "handle" other than the object itself. You must instantiate an instance of the persistent class itself and populate its identifier properties before you can load() the persistent state associated with a composite key. We call this approach an embedded composite identifier, and discourage it for serious applications.

A second approach is what we call a mapped composite identifier, where the identifier properties named inside the <composite-id> element are duplicated on both the persistent class and a separate identifier class.

<composite-id class="MedicareId" mapped="true">
        <key-property name="medicareNumber"/>
        <key-property name="dependent"/>
</composite-id>

In this example, both the composite identifier class, MedicareId, and the entity class itself have properties named medicareNumber and dependent. The identifier class must override equals() and hashCode() and implement. Serializable. The disadvantage of this approach is quite obvious—code duplication.

The following attributes are used to specify a mapped composite identifier:

We will describe a third, even more convenient approach where the composite identifier is implemented as a component class in Section 8.4, “Components as composite identifiers”. The attributes described below apply only to this alternative approach:

This third approach, an identifier component is the one we recommend for almost all applications.

The <discriminator> element is required for polymorphic persistence using the table-per-class-hierarchy mapping strategy and declares a discriminator column of the table. The discriminator column contains marker values that tell the persistence layer what subclass to instantiate for a particular row. A restricted set of types may be used: string, character, integer, byte, short, boolean, yes_no, true_false.

<discriminator
        column="discriminator_column"                      (1)
        type="discriminator_type"                          (2)
        force="true|false"                                 (3)
        insert="true|false"                                (4)
        formula="arbitrary sql expression"                 (5)
/>

Actual values of the discriminator column are specified by the discriminator-value attribute of the <class> and <subclass> elements.

The force attribute is (only) useful if the table contains rows with "extra" discriminator values that are not mapped to a persistent class. This will not usually be the case.

Using the formula attribute you can declare an arbitrary SQL expression that will be used to evaluate the type of a row:

<discriminator
    formula="case when CLASS_TYPE in ('a', 'b', 'c') then 0 else 1 end"
    type="integer"/>

The <version> element is optional and indicates that the table contains versioned data. This is particularly useful if you plan to use long transactions (see below).

<version
        column="version_column"                                      (1)
        name="propertyName"                                          (2)
        type="typename"                                              (3)
        access="field|property|ClassName"                            (4)
        unsaved-value="null|negative|undefined"                      (5)
        generated="never|always"                                     (6)
        insert="true|false"                                          (7)
        node="element-name|@attribute-name|element/@attribute|."
/>

Version numbers may be of Hibernate type long, integer, short, timestamp or calendar.

A version or timestamp property should never be null for a detached instance, so Hibernate will detact any instance with a null version or timestamp as transient, no matter what other unsaved-value strategies are specified. Declaring a nullable version or timestamp property is an easy way to avoid any problems with transitive reattachment in Hibernate, especially useful for people using assigned identifiers or composite keys!

The optional <timestamp> element indicates that the table contains timestamped data. This is intended as an alternative to versioning. Timestamps are by nature a less safe implementation of optimistic locking. However, sometimes the application might use the timestamps in other ways.

<timestamp
        column="timestamp_column"                                    (1)
        name="propertyName"                                          (2)
        access="field|property|ClassName"                            (3)
        unsaved-value="null|undefined"                               (4)
        source="vm|db"                                               (5)
        generated="never|always"                                     (6)
        node="element-name|@attribute-name|element/@attribute|."
/>

Note that <timestamp> is equivalent to <version type="timestamp">. And <timestamp source="db"> is equivalent to <version type="dbtimestamp">

The <property> element declares a persistent, JavaBean style property of the class.

<property
        name="propertyName"                                          (1)
        column="column_name"                                         (2)
        type="typename"                                              (3)
        update="true|false"                                          (4)
        insert="true|false"                                          (4)
        formula="arbitrary SQL expression"                           (5)
        access="field|property|ClassName"                            (6)
        lazy="true|false"                                            (7)
        unique="true|false"                                          (8)
        not-null="true|false"                                        (9)
        optimistic-lock="true|false"                                 (10)
        generated="never|insert|always"                              (11)
        node="element-name|@attribute-name|element/@attribute|."
        index="index_name"
        unique_key="unique_key_id"
        length="L"
        precision="P"
        scale="S"
/>

typename could be:

If you do not specify a type, Hibernate will use reflection upon the named property to take a guess at the correct Hibernate type. Hibernate will try to interpret the name of the return class of the property getter using rules 2, 3, 4 in that order. However, this is not always enough. In certain cases you will still need the type attribute. (For example, to distinguish between Hibernate.DATE and Hibernate.TIMESTAMP, or to specify a custom type.)

The access attribute lets you control how Hibernate will access the property at runtime. By default, Hibernate will call the property get/set pair. If you specify access="field", Hibernate will bypass the get/set pair and access the field directly, using reflection. You may specify your own strategy for property access by naming a class that implements the interface org.hibernate.property.PropertyAccessor.

An especially powerful feature are derived properties. These properties are by definition read-only, the property value is computed at load time. You declare the computation as a SQL expression, this translates to a SELECT clause subquery in the SQL query that loads an instance:

<property name="totalPrice"
    formula="( SELECT SUM (li.quantity*p.price) FROM LineItem li, Product p
                WHERE li.productId = p.productId
                AND li.customerId = customerId
                AND li.orderNumber = orderNumber )"/>

Note that you can reference the entities own table by not declaring an alias on a particular column (customerId in the given example). Also note that you can use the nested <formula> mapping element if you don't like to use the attribute.

An ordinary association to another persistent class is declared using a many-to-one element. The relational model is a many-to-one association: a foreign key in one table is referencing the primary key column(s) of the target table.

<many-to-one
        name="propertyName"                                          (1)
        column="column_name"                                         (2)
        class="ClassName"                                            (3)
        cascade="cascade_style"                                      (4)
        fetch="join|select"                                          (5)
        update="true|false"                                          (6)
        insert="true|false"                                          (6)
        property-ref="propertyNameFromAssociatedClass"               (7)
        access="field|property|ClassName"                            (8)
        unique="true|false"                                          (9)
        not-null="true|false"                                        (10)
        optimistic-lock="true|false"                                 (11)
        lazy="proxy|no-proxy|false"                                  (12)
        not-found="ignore|exception"                                 (13)
        entity-name="EntityName"                                     (14)
        formula="arbitrary SQL expression"                           (15)
        node="element-name|@attribute-name|element/@attribute|."
        embed-xml="true|false"
        index="index_name"
        unique_key="unique_key_id"
        foreign-key="foreign_key_name"
/>

Setting a value of the cascade attribute to any meaningful value other than none will propagate certain operations to the associated object. The meaningful values are the names of Hibernate's basic operations, persist, merge, delete, save-update, evict, replicate, lock, refresh, as well as the special values delete-orphan and all and comma-separated combinations of operation names, for example, cascade="persist,merge,evict" or cascade="all,delete-orphan". See Section 10.11, “Transitive persistence” for a full explanation. Note that single valued associations (many-to-one and one-to-one associations) do not support orphan delete.

A typical many-to-one declaration looks as simple as this:

<many-to-one name="product" class="Product" column="PRODUCT_ID"/>

The property-ref attribute should only be used for mapping legacy data where a foreign key refers to a unique key of the associated table other than the primary key. This is an ugly relational model. For example, suppose the Product class had a unique serial number, that is not the primary key. (The unique attribute controls Hibernate's DDL generation with the SchemaExport tool.)

<property name="serialNumber" unique="true" type="string" column="SERIAL_NUMBER"/>

Then the mapping for OrderItem might use:

<many-to-one name="product" property-ref="serialNumber" column="PRODUCT_SERIAL_NUMBER"/>

This is certainly not encouraged, however.

If the referenced unique key comprises multiple properties of the associated entity, you should map the referenced properties inside a named <properties> element.

If the referenced unique key is the property of a component, you may specify a property path:

<many-to-one name="owner" property-ref="identity.ssn" column="OWNER_SSN"/>

A one-to-one association to another persistent class is declared using a one-to-one element.

<one-to-one
        name="propertyName"                                          (1)
        class="ClassName"                                            (2)
        cascade="cascade_style"                                      (3)
        constrained="true|false"                                     (4)
        fetch="join|select"                                          (5)
        property-ref="propertyNameFromAssociatedClass"               (6)
        access="field|property|ClassName"                            (7)
        formula="any SQL expression"                                 (8)
        lazy="proxy|no-proxy|false"                                  (9)
        entity-name="EntityName"                                     (10)
        node="element-name|@attribute-name|element/@attribute|."
        embed-xml="true|false"
        foreign-key="foreign_key_name"
/>

There are two varieties of one-to-one association:

Primary key associations don't need an extra table column; if two rows are related by the association then the two table rows share the same primary key value. So if you want two objects to be related by a primary key association, you must make sure that they are assigned the same identifier value!

For a primary key association, add the following mappings to Employee and Person, respectively.

<one-to-one name="person" class="Person"/>

<one-to-one name="employee" class="Employee" constrained="true"/>

Now we must ensure that the primary keys of related rows in the PERSON and EMPLOYEE tables are equal. We use a special Hibernate identifier generation strategy called foreign:

<class name="person" table="PERSON">
    <id name="id" column="PERSON_ID">
        <generator class="foreign">
            <param name="property">employee</param>
        </generator>
    </id>
    ...
    <one-to-one name="employee"
        class="Employee"
        constrained="true"/>
</class>

A newly saved instance of Person is then assigned the same primary key value as the Employee instance refered with the employee property of that Person.

Alternatively, a foreign key with a unique constraint, from Employee to Person, may be expressed as:

<many-to-one name="person" class="Person" column="PERSON_ID" unique="true"/>

And this association may be made bidirectional by adding the following to the Person mapping:

<one-to-one name"employee" class="Employee" property-ref="person"/>

<natural-id mutable="true|false"/>
        <property ... />
        <many-to-one ... />
        ......
</natural-id>

Even though we recommend the use of surrogate keys as primary keys, you should still try to identify natural keys for all entities. A natural key is a property or combination of properties that is unique and non-null. If it is also immutable, even better. Map the properties of the natural key inside the <natural-id> element. Hibernate will generate the necessary unique key and nullability constraints, and your mapping will be more self-documenting.

We strongly recommend that you implement equals() and hashCode() to compare the natural key properties of the entity.

This mapping is not intended for use with entities with natural primary keys.

The <component> element maps properties of a child object to columns of the table of a parent class. Components may, in turn, declare their own properties, components or collections. See "Components" below.

<component 
        name="propertyName"                 (1)
        class="className"                   (2)
        insert="true|false"                 (3)
        update="true|false"                 (4)
        access="field|property|ClassName"   (5)
        lazy="true|false"                   (6)
        optimistic-lock="true|false"        (7)
        unique="true|false"                 (8)
        node="element-name|."
>
        
        <property ...../>
        <many-to-one .... />
        ........
</component>

The child <property> tags map properties of the child class to table columns.

The <component> element allows a <parent> subelement that maps a property of the component class as a reference back to the containing entity.

The <dynamic-component> element allows a Map to be mapped as a component, where the property names refer to keys of the map, see Section 8.5, “Dynamic components”.

The <properties> element allows the definition of a named, logical grouping of properties of a class. The most important use of the construct is that it allows a combination of properties to be the target of a property-ref. It is also a convenient way to define a multi-column unique constraint.

<properties 
        name="logicalName"                  (1)
        insert="true|false"                 (2)
        update="true|false"                 (3)
        optimistic-lock="true|false"        (4)
        unique="true|false"                 (5)
>
        
        <property ...../>
        <many-to-one .... />
        ........
</properties>

For example, if we have the following <properties> mapping:

<class name="Person">
    <id name="personNumber"/>
    ...
    <properties name="name" 
            unique="true" update="false">
        <property name="firstName"/>
        <property name="initial"/>
        <property name="lastName"/>
    </properties>
</class>

Then we might have some legacy data association which refers to this unique key of the Person table, instead of to the primary key:

<many-to-one name="person" 
         class="Person" property-ref="name">
    <column name="firstName"/>
    <column name="initial"/>
    <column name="lastName"/>
</many-to-one>

We don't recommend the use of this kind of thing outside the context of mapping legacy data.

Finally, polymorphic persistence requires the declaration of each subclass of the root persistent class. For the table-per-class-hierarchy mapping strategy, the <subclass> declaration is used.

<subclass
        name="ClassName"                              (1)
        discriminator-value="discriminator_value"     (2)
        proxy="ProxyInterface"                        (3)
        lazy="true|false"                             (4)
        dynamic-update="true|false"
        dynamic-insert="true|false"
        entity-name="EntityName"
        node="element-name"
        extends="SuperclassName">

        <property .... />
        .....
</subclass>

Each subclass should declare its own persistent properties and subclasses. <version> and <id> properties are assumed to be inherited from the root class. Each subclass in a heirarchy must define a unique discriminator-value. If none is specified, the fully qualified Java class name is used.

For information about inheritance mappings, see Chapter 9, Inheritance Mapping.

Alternatively, each subclass may be mapped to its own table (table-per-subclass mapping strategy). Inherited state is retrieved by joining with the table of the superclass. We use the <joined-subclass> element.

<joined-subclass
        name="ClassName"                    (1)
        table="tablename"                   (2)
        proxy="ProxyInterface"              (3)
        lazy="true|false"                   (4)
        dynamic-update="true|false"
        dynamic-insert="true|false"
        schema="schema"
        catalog="catalog"
        extends="SuperclassName"
        persister="ClassName"
        subselect="SQL expression"
        entity-name="EntityName"
        node="element-name">

        <key .... >

        <property .... />
        .....
</joined-subclass>

No discriminator column is required for this mapping strategy. Each subclass must, however, declare a table column holding the object identifier using the <key> element. The mapping at the start of the chapter would be re-written as:

<?xml version="1.0"?>
<!DOCTYPE hibernate-mapping PUBLIC
        "-//Hibernate/Hibernate Mapping DTD//EN"
        "http://hibernate.sourceforge.net/hibernate-mapping-3.0.dtd">

<hibernate-mapping package="eg">

        <class name="Cat" table="CATS">
                <id name="id" column="uid" type="long">
                        <generator class="hilo"/>
                </id>
                <property name="birthdate" type="date"/>
                <property name="color" not-null="true"/>
                <property name="sex" not-null="true"/>
                <property name="weight"/>
                <many-to-one name="mate"/>
                <set name="kittens">
                        <key column="MOTHER"/>
                        <one-to-many class="Cat"/>
                </set>
                <joined-subclass name="DomesticCat" table="DOMESTIC_CATS">
                    <key column="CAT"/>
                    <property name="name" type="string"/>
                </joined-subclass>
        </class>

        <class name="eg.Dog">
                <!-- mapping for Dog could go here -->
        </class>

</hibernate-mapping>

For information about inheritance mappings, see Chapter 9, Inheritance Mapping.

A third option is to map only the concrete classes of an inheritance hierarchy to tables, (the table-per-concrete-class strategy) where each table defines all persistent state of the class, including inherited state. In Hibernate, it is not absolutely necessary to explicitly map such inheritance hierarchies. You can simply map each class with a separate <class> declaration. However, if you wish use polymorphic associations (e.g. an association to the superclass of your hierarchy), you need to use the <union-subclass> mapping.

<union-subclass
        name="ClassName"                    (1)
        table="tablename"                   (2)
        proxy="ProxyInterface"              (3)
        lazy="true|false"                   (4)
        dynamic-update="true|false"
        dynamic-insert="true|false"
        schema="schema"
        catalog="catalog"
        extends="SuperclassName"
        abstract="true|false"
        persister="ClassName"
        subselect="SQL expression"
        entity-name="EntityName"
        node="element-name">

        <property .... />
        .....
</union-subclass>

No discriminator column or key column is required for this mapping strategy.

For information about inheritance mappings, see Chapter 9, Inheritance Mapping.

Using the <join> element, it is possible to map properties of one class to several tables.

<join
        table="tablename"                        (1)
        schema="owner"                           (2)
        catalog="catalog"                        (3)
        fetch="join|select"                      (4)
        inverse="true|false"                     (5)
        optional="true|false">                   (6)
        
        <key ... />
        
        <property ... />
        ...
</join>

For example, the address information for a person can be mapped to a separate table (while preserving value type semantics for all properties):

<class name="Person"
    table="PERSON">

    <id name="id" column="PERSON_ID">...</id>

    <join table="ADDRESS">
        <key column="ADDRESS_ID"/>
        <property name="address"/>
        <property name="zip"/>
        <property name="country"/>
    </join>
    ...

This feature is often only useful for legacy data models, we recommend fewer tables than classes and a fine-grained domain model. However, it is useful for switching between inheritance mapping strategies in a single hierarchy, as explained later.

We've seen the <key> element crop up a few times now. It appears anywhere the parent mapping element defines a join to a new table, and defines the foreign key in the joined table, that references the primary key of the original table.

<key
        column="columnname"                      (1)
        on-delete="noaction|cascade"             (2)
        property-ref="propertyName"              (3)
        not-null="true|false"                    (4)
        update="true|false"                      (5)
        unique="true|false"                      (6)
/>

We recommend that for systems where delete performance is important, all keys should be defined on-delete="cascade", and Hibernate will use a database-level ON CASCADE DELETE constraint, instead of many individual DELETE statements. Be aware that this feature bypasses Hibernate's usual optimistic locking strategy for versioned data.

The not-null and update attributes are useful when mapping a unidirectional one to many association. If you map a unidirectional one to many to a non-nullable foreign key, you must declare the key column using <key not-null="true">.

Any mapping element which accepts a column attribute will alternatively accept a <column> subelement. Likewise, <formula> is an alternative to the formula attribute.

<column
        name="column_name"
        length="N"
        precision="N"
        scale="N"
        not-null="true|false"
        unique="true|false"
        unique-key="multicolumn_unique_key_name"
        index="index_name"
        sql-type="sql_type_name"
        check="SQL expression"
        default="SQL expression"/>

<formula>SQL expression</formula>

column and formula attributes may even be combined within the same property or association mapping to express, for example, exotic join conditions.

<many-to-one name="homeAddress" class="Address"
        insert="false" update="false">
    <column name="person_id" not-null="true" length="10"/>
    <formula>'MAILING'</formula>
</many-to-one>

Suppose your application has two persistent classes with the same name, and you don't want to specify the fully qualified (package) name in Hibernate queries. Classes may be "imported" explicitly, rather than relying upon auto-import="true". You may even import classes and interfaces that are not explicitly mapped.

<import class="java.lang.Object" rename="Universe"/>

<import
        class="ClassName"              (1)
        rename="ShortName"             (2)
/>

There is one further type of property mapping. The <any> mapping element defines a polymorphic association to classes from multiple tables. This type of mapping always requires more than one column. The first column holds the type of the associated entity. The remaining columns hold the identifier. It is impossible to specify a foreign key constraint for this kind of association, so this is most certainly not meant as the usual way of mapping (polymorphic) associations. You should use this only in very special cases (eg. audit logs, user session data, etc).

The meta-type attribute lets the application specify a custom type that maps database column values to persistent classes which have identifier properties of the type specified by id-type. You must specify the mapping from values of the meta-type to class names.

<any name="being" id-type="long" meta-type="string">
    <meta-value value="TBL_ANIMAL" class="Animal"/>
    <meta-value value="TBL_HUMAN" class="Human"/>
    <meta-value value="TBL_ALIEN" class="Alien"/>
    <column name="table_name"/>
    <column name="id"/>
</any>

<any
        name="propertyName"                      (1)
        id-type="idtypename"                     (2)
        meta-type="metatypename"                 (3)
        cascade="cascade_style"                  (4)
        access="field|property|ClassName"        (5)
        optimistic-lock="true|false"             (6)
>
        <meta-value ... />
        <meta-value ... />
        .....
        <column .... />
        <column .... />
        .....
</any>

To understand the behaviour of various Java language-level objects with respect to the persistence service, we need to classify them into two groups:

An entity exists independently of any other objects holding references to the entity. Contrast this with the usual Java model where an unreferenced object is garbage collected. Entities must be explicitly saved and deleted (except that saves and deletions may be cascaded from a parent entity to its children). This is different from the ODMG model of object persistence by reachablity - and corresponds more closely to how application objects are usually used in large systems. Entities support circular and shared references. They may also be versioned.

An entity's persistent state consists of references to other entities and instances of value types. Values are primitives, collections (not what's inside a collection), components and certain immutable objects. Unlike entities, values (in particular collections and components) are persisted and deleted by reachability. Since value objects (and primitives) are persisted and deleted along with their containing entity they may not be independently versioned. Values have no independent identity, so they cannot be shared by two entities or collections.

Up until now, we've been using the term "persistent class" to refer to entities. We will continue to do that. Strictly speaking, however, not all user-defined classes with persistent state are entities. A component is a user defined class with value semantics. A Java property of type java.lang.String also has value semantics. Given this definition, we can say that all types (classes) provided by the JDK have value type semantics in Java, while user-defined types may be mapped with entity or value type semantics. This decision is up to the application developer. A good hint for an entity class in a domain model are shared references to a single instance of that class, while composition or aggregation usually translates to a value type.

We'll revisit both concepts throughout the documentation.

The challenge is to map the Java type system (and the developers' definition of entities and value types) to the SQL/database type system. The bridge between both systems is provided by Hibernate: for entities we use <class>, <subclass> and so on. For value types we use <property>, <component>, etc, usually with a type attribute. The value of this attribute is the name of a Hibernate mapping type. Hibernate provides many mappings (for standard JDK value types) out of the box. You can write your own mapping types and implement your custom conversion strategies as well, as you'll see later.

All built-in Hibernate types except collections support null semantics.

The built-in basic mapping types may be roughly categorized into

Unique identifiers of entities and collections may be of any basic type except binary, blob and clob. (Composite identifiers are also allowed, see below.)

The basic value types have corresponding Type constants defined on org.hibernate.Hibernate. For example, Hibernate.STRING represents the string type.

It is relatively easy for developers to create their own value types. For example, you might want to persist properties of type java.lang.BigInteger to VARCHAR columns. Hibernate does not provide a built-in type for this. But custom types are not limited to mapping a property (or collection element) to a single table column. So, for example, you might have a Java property getName()/setName() of type java.lang.String that is persisted to the columns FIRST_NAME, INITIAL, SURNAME.

To implement a custom type, implement either org.hibernate.UserType or org.hibernate.CompositeUserType and declare properties using the fully qualified classname of the type. Check out org.hibernate.test.DoubleStringType to see the kind of things that are possible.

<property name="twoStrings" type="org.hibernate.test.DoubleStringType">
    <column name="first_string"/>
    <column name="second_string"/>
</property>

Notice the use of <column> tags to map a property to multiple columns.

The CompositeUserType, EnhancedUserType, UserCollectionType, and UserVersionType interfaces provide support for more specialized uses.

You may even supply parameters to a UserType in the mapping file. To do this, your UserType must implement the org.hibernate.usertype.ParameterizedType interface. To supply parameters to your custom type, you can use the <type> element in your mapping files.

<property name="priority">
    <type name="com.mycompany.usertypes.DefaultValueIntegerType">
        <param name="default">0</param>
    </type>
</property>

The UserType can now retrieve the value for the parameter named default from the Properties object passed to it.

If you use a certain UserType very often, it may be useful to define a shorter name for it. You can do this using the <typedef> element. Typedefs assign a name to a custom type, and may also contain a list of default parameter values if the type is parameterized.

<typedef class="com.mycompany.usertypes.DefaultValueIntegerType" name="default_zero">
    <param name="default">0</param>
</typedef>

<property name="priority" type="default_zero"/>

It is also possible to override the parameters supplied in a typedef on a case-by-case basis by using type parameters on the property mapping.

Even though Hibernate's rich range of built-in types and support for components means you will very rarely need to use a custom type, it is nevertheless considered good form to use custom types for (non-entity) classes that occur frequently in your application. For example, a MonetaryAmount class is a good candidate for a CompositeUserType, even though it could easily be mapped as a component. One motivation for this is abstraction. With a custom type, your mapping documents would be future-proofed against possible changes in your way of representing monetary values.

Many Hibernate users prefer to embed mapping information directly in sourcecode using XDoclet @hibernate.tags. We will not cover this approach in this document, since strictly it is considered part of XDoclet. However, we include the following example of the Cat class with XDoclet mappings.

package eg;
import java.util.Set;
import java.util.Date;

/**
 * @hibernate.class
 *  table="CATS"
 */
public class Cat {
    private Long id; // identifier
    private Date birthdate;
    private Cat mother;
    private Set kittens
    private Color color;
    private char sex;
    private float weight;

    /*
     * @hibernate.id
     *  generator-class="native"
     *  column="CAT_ID"
     */
    public Long getId() {
        return id;
    }
    private void setId(Long id) {
        this.id=id;
    }

    /**
     * @hibernate.many-to-one
     *  column="PARENT_ID"
     */
    public Cat getMother() {
        return mother;
    }
    void setMother(Cat mother) {
        this.mother = mother;
    }

    /**
     * @hibernate.property
     *  column="BIRTH_DATE"
     */
    public Date getBirthdate() {
        return birthdate;
    }
    void setBirthdate(Date date) {
        birthdate = date;
    }
    /**
     * @hibernate.property
     *  column="WEIGHT"
     */
    public float getWeight() {
        return weight;
    }
    void setWeight(float weight) {
        this.weight = weight;
    }

    /**
     * @hibernate.property
     *  column="COLOR"
     *  not-null="true"
     */
    public Color getColor() {
        return color;
    }
    void setColor(Color color) {
        this.color = color;
    }
    /**
     * @hibernate.set
     *  inverse="true"
     *  order-by="BIRTH_DATE"
     * @hibernate.collection-key
     *  column="PARENT_ID"
     * @hibernate.collection-one-to-many
     */
    public Set getKittens() {
        return kittens;
    }
    void setKittens(Set kittens) {
        this.kittens = kittens;
    }
    // addKitten not needed by Hibernate
    public void addKitten(Cat kitten) {
        kittens.add(kitten);
    }

    /**
     * @hibernate.property
     *  column="SEX"
     *  not-null="true"
     *  update="false"
     */
    public char getSex() {
        return sex;
    }
    void setSex(char sex) {
        this.sex=sex;
    }
}

See the Hibernate web site for more examples of XDoclet and Hibernate.

JDK 5.0 introduced XDoclet-style annotations at the language level, type-safe and checked at compile time. This mechnism is more powerful than XDoclet annotations and better supported by tools and IDEs. IntelliJ IDEA, for example, supports auto-completion and syntax highlighting of JDK 5.0 annotations. The new revision of the EJB specification (JSR-220) uses JDK 5.0 annotations as the primary metadata mechanism for entity beans. Hibernate3 implements the EntityManager of JSR-220 (the persistence API), support for mapping metadata is available via the Hibernate Annotations package, as a separate download. Both EJB3 (JSR-220) and Hibernate3 metadata is supported.

This is an example of a POJO class annotated as an EJB entity bean:

@Entity(access = AccessType.FIELD)
public class Customer implements Serializable {

    @Id;
    Long id;

    String firstName;
    String lastName;
    Date birthday;

    @Transient
    Integer age;

    @Embedded
    private Address homeAddress;

    @OneToMany(cascade=CascadeType.ALL)
    @JoinColumn(name="CUSTOMER_ID")
    Set<Order> orders;

    // Getter/setter and business methods
}

Note that support for JDK 5.0 Annotations (and JSR-220) is still work in progress and not completed. Please refer to the Hibernate Annotations module for more details.

Collection instances are distinguished in the database by the foreign key of the entity that owns the collection. This foreign key is referred to as the collection key column (or columns) of the collection table. The collection key column is mapped by the <key> element.

There may be a nullability constraint on the foreign key column. For most collections, this is implied. For unidirectional one to many associations, the foreign key column is nullable by default, so you might need to specify not-null="true".

<key column="productSerialNumber" not-null="true"/>

The foreign key constraint may use ON DELETE CASCADE.

<key column="productSerialNumber" on-delete="cascade"/>

See the previous chapter for a full definition of the <key> element.

Collections may contain almost any other Hibernate type, including all basic types, custom types, components, and of course, references to other entities. This is an important distinction: an object in a collection might be handled with "value" semantics (its lifecycle fully depends on the collection owner) or it might be a reference to another entity, with its own lifecycle. In the latter case, only the "link" between the two objects is considered to be state held by the collection.

The contained type is referred to as the collection element type. Collection elements are mapped by <element> or <composite-element>, or in the case of entity references, with <one-to-many> or <many-to-many>. The first two map elements with value semantics, the next two are used to map entity associations.

All collection mappings, except those with set and bag semantics, need an index column in the collection table - a column that maps to an array index, or List index, or Map key. The index of a Map may be of any basic type, mapped with <map-key>, it may be an entity reference mapped with <map-key-many-to-many>, or it may be a composite type, mapped with <composite-map-key>. The index of an array or list is always of type integer and is mapped using the <list-index> element. The mapped column contains sequential integers (numbered from zero, by default).

<list-index 
        column="column_name"                (1)
        base="0|1|..."/>

<map-key 
        column="column_name"                (1)
        formula="any SQL expression"        (2)
        type="type_name"                    (3)
        node="@attribute-name"
        length="N"/>

<map-key-many-to-many
        column="column_name"                (1)
        formula="any SQL expression"        (2)(3)
        class="ClassName"
/>

If your table doesn't have an index column, and you still wish to use List as the property type, you should map the property as a Hibernate <bag>. A bag does not retain its order when it is retrieved from the database, but it may be optionally sorted or ordered.

Any collection of values or many-to-many association requires a dedicated collection table with a foreign key column or columns, collection element column or columns and possibly an index column or columns.

For a collection of values, we use the <element> tag.

<element
        column="column_name"                     (1)
        formula="any SQL expression"             (2)
        type="typename"                          (3)
        length="L"
        precision="P"
        scale="S"
        not-null="true|false"
        unique="true|false"
        node="element-name"
/>

A many-to-many association is specified using the <many-to-many> element.

<many-to-many
        column="column_name"                               (1)
        formula="any SQL expression"                       (2)
        class="ClassName"                                  (3)
        fetch="select|join"                                (4)
        unique="true|false"                                (5)
        not-found="ignore|exception"                       (6)
        entity-name="EntityName"                           (7)
        property-ref="propertyNameFromAssociatedClass"     (8)
        node="element-name"
        embed-xml="true|false"
    />

Some examples, first, a set of strings:

<set name="names" table="person_names">
    <key column="person_id"/>
    <element column="person_name" type="string"/>
</set>

A bag containing integers (with an iteration order determined by the order-by attribute):

<bag name="sizes" 
        table="item_sizes" 
        order-by="size asc">
    <key column="item_id"/>
    <element column="size" type="integer"/>
</bag>

An array of entities - in this case, a many to many association:

<array name="addresses" 
        table="PersonAddress" 
        cascade="persist">
    <key column="personId"/>
    <list-index column="sortOrder"/>
    <many-to-many column="addressId" class="Address"/>
</array>

A map from string indices to dates:

<map name="holidays" 
        table="holidays" 
        schema="dbo" 
        order-by="hol_name asc">
    <key column="id"/>
    <map-key column="hol_name" type="string"/>
    <element column="hol_date" type="date"/>
</map>

A list of components (discussed in the next chapter):

<list name="carComponents" 
        table="CarComponents">
    <key column="carId"/>
    <list-index column="sortOrder"/>
    <composite-element class="CarComponent">
        <property name="price"/>
        <property name="type"/>
        <property name="serialNumber" column="serialNum"/>
    </composite-element>
</list>

A one to many association links the tables of two classes via a foreign key, with no intervening collection table. This mapping loses certain semantics of normal Java collections:

An association from Product to Part requires existence of a foreign key column and possibly an index column to the Part table. A <one-to-many> tag indicates that this is a one to many association.

<one-to-many 
        class="ClassName"                                  (1)
        not-found="ignore|exception"                       (2)
        entity-name="EntityName"                           (3)
        node="element-name"
        embed-xml="true|false"
    />

Notice that the <one-to-many> element does not need to declare any columns. Nor is it necessary to specify the table name anywhere.

Very important note: If the foreign key column of a <one-to-many> association is declared NOT NULL, you must declare the <key> mapping not-null="true" or use a bidirectional association with the collection mapping marked inverse="true". See the discussion of bidirectional associations later in this chapter.

This example shows a map of Part entities by name (where partName is a persistent property of Part). Notice the use of a formula-based index.

<map name="parts"
        cascade="all">
    <key column="productId" not-null="true"/>
    <map-key formula="partName"/>
    <one-to-many class="Part"/>
</map>

Hibernate supports collections implementing java.util.SortedMap and java.util.SortedSet. You must specify a comparator in the mapping file:

<set name="aliases" 
            table="person_aliases" 
            sort="natural">
    <key column="person"/>
    <element column="name" type="string"/>
</set>

<map name="holidays" sort="my.custom.HolidayComparator">
    <key column="year_id"/>
    <map-key column="hol_name" type="string"/>
    <element column="hol_date" type="date"/>
</map>

Allowed values of the sort attribute are unsorted, natural and the name of a class implementing java.util.Comparator.

Sorted collections actually behave like java.util.TreeSet or java.util.TreeMap.

If you want the database itself to order the collection elements use the order-by attribute of set, bag or map mappings. This solution is only available under JDK 1.4 or higher (it is implemented using LinkedHashSet or LinkedHashMap). This performs the ordering in the SQL query, not in memory.

<set name="aliases" table="person_aliases" order-by="lower(name) asc">
    <key column="person"/>
    <element column="name" type="string"/>
</set>

<map name="holidays" order-by="hol_date, hol_name">
    <key column="year_id"/>
    <map-key column="hol_name" type="string"/>
    <element column="hol_date type="date"/>
</map>

Note that the value of the order-by attribute is an SQL ordering, not a HQL ordering!

Associations may even be sorted by some arbitrary criteria at runtime using a collection filter().

sortedUsers = s.createFilter( group.getUsers(), "order by this.name" ).list();

A bidirectional association allows navigation from both "ends" of the association. Two kinds of bidirectional association are supported:

You may specify a bidirectional many-to-many association simply by mapping two many-to-many associations to the same database table and declaring one end as inverse (which one is your choice, but it can not be an indexed collection).

Here's an example of a bidirectional many-to-many association; each category can have many items and each item can be in many categories:

<class name="Category">
    <id name="id" column="CATEGORY_ID"/>
    ...
    <bag name="items" table="CATEGORY_ITEM">
        <key column="CATEGORY_ID"/>
        <many-to-many class="Item" column="ITEM_ID"/>
    </bag>
</class>

<class name="Item">
    <id name="id" column="CATEGORY_ID"/>
    ...

    <!-- inverse end -->
    <bag name="categories" table="CATEGORY_ITEM" inverse="true">
        <key column="ITEM_ID"/>
        <many-to-many class="Category" column="CATEGORY_ID"/>
    </bag>
</class>

Changes made only to the inverse end of the association are not persisted. This means that Hibernate has two representations in memory for every bidirectional association, one link from A to B and another link from B to A. This is easier to understand if you think about the Java object model and how we create a many-to-many relationship in Java:

category.getItems().add(item);          // The category now "knows" about the relationship
item.getCategories().add(category);     // The item now "knows" about the relationship

session.persist(item);                   // The relationship won't be saved!
session.persist(category);               // The relationship will be saved

The non-inverse side is used to save the in-memory representation to the database.

You may define a bidirectional one-to-many association by mapping a one-to-many association to the same table column(s) as a many-to-one association and declaring the many-valued end inverse="true".

<class name="Parent">
    <id name="id" column="parent_id"/>
    ....
    <set name="children" inverse="true">
        <key column="parent_id"/>
        <one-to-many class="Child"/>
    </set>
</class>

<class name="Child">
    <id name="id" column="child_id"/>
    ....
    <many-to-one name="parent" 
        class="Parent" 
        column="parent_id"
        not-null="true"/>
</class>

Mapping one end of an association with inverse="true" doesn't affect the operation of cascades, these are orthogonal concepts!

A bidirectional association where one end is represented as a <list> or <map> requires special consideration. If there is a property of the child class which maps to the index column, no problem, we can continue using inverse="true" on the collection mapping:

<class name="Parent">
    <id name="id" column="parent_id"/>
    ....
    <map name="children" inverse="true">
        <key column="parent_id"/>
        <map-key column="name" 
            type="string"/>
        <one-to-many class="Child"/>
    </map>
</class>

<class name="Child">
    <id name="id" column="child_id"/>
    ....
    <property name="name" 
        not-null="true"/>
    <many-to-one name="parent" 
        class="Parent" 
        column="parent_id"
        not-null="true"/>
</class>

But, if there is no such property on the child class, we can't think of the association as truly bidirectional (there is information available at one end of the association that is not available at the other end). In this case, we can't map the collection inverse="true". Instead, we could use the following mapping:

<class name="Parent">
    <id name="id" column="parent_id"/>
    ....
    <map name="children">
        <key column="parent_id"
            not-null="true"/>
        <map-key column="name" 
            type="string"/>
        <one-to-many class="Child"/>
    </map>
</class>

<class name="Child">
    <id name="id" column="child_id"/>
    ....
    <many-to-one name="parent" 
        class="Parent" 
        column="parent_id"
        insert="false"
        update="false"
        not-null="true"/>
</class>

Note that in this mapping, the collection-valued end of the association is responsible for updates to the foreign key. TODO: Does this really result in some unnecessary update statements?

There are three possible approaches to mapping a ternary association. One is to use a Map with an association as its index:

<map name="contracts">
    <key column="employer_id" not-null="true"/>
    <map-key-many-to-many column="employee_id" class="Employee"/>
    <one-to-many class="Contract"/>
</map>

<map name="connections">
    <key column="incoming_node_id"/>
    <map-key-many-to-many column="outgoing_node_id" class="Node"/>
    <many-to-many column="connection_id" class="Connection"/>
</map>

A second approach is to simply remodel the association as an entity class. This is the approach we use most commonly.

A final alternative is to use composite elements, which we will discuss later.

If you've fully embraced our view that composite keys are a bad thing and that entities should have synthetic identifiers (surrogate keys), then you might find it a bit odd that the many to many associations and collections of values that we've shown so far all map to tables with composite keys! Now, this point is quite arguable; a pure association table doesn't seem to benefit much from a surrogate key (though a collection of composite values might). Nevertheless, Hibernate provides a feature that allows you to map many to many associations and collections of values to a table with a surrogate key.

The <idbag> element lets you map a List (or Collection) with bag semantics.

<idbag name="lovers" table="LOVERS">
    <collection-id column="ID" type="long">
        <generator class="sequence"/>
    </collection-id>
    <key column="PERSON1"/>
    <many-to-many column="PERSON2" class="Person" fetch="join"/>
</idbag>

As you can see, an <idbag> has a synthetic id generator, just like an entity class! A different surrogate key is assigned to each collection row. Hibernate does not provide any mechanism to discover the surrogate key value of a particular row, however.

Note that the update performance of an <idbag> is much better than a regular <bag>! Hibernate can locate individual rows efficiently and update or delete them individually, just like a list, map or set.

In the current implementation, the native identifier generation strategy is not supported for <idbag> collection identifiers.

A unidirectional many-to-one association is the most common kind of unidirectional association.

<class name="Person">
    <id name="id" column="personId">
        <generator class="native"/>
    </id>
    <many-to-one name="address" 
        column="addressId"
        not-null="true"/>
</class>

<class name="Address">
    <id name="id" column="addressId">
        <generator class="native"/>
    </id>
</class>

create table Person ( personId bigint not null primary key, addressId bigint not null )
create table Address ( addressId bigint not null primary key )
        

A unidirectional one-to-one association on a foreign key is almost identical. The only difference is the column unique constraint.

<class name="Person">
    <id name="id" column="personId">
        <generator class="native"/>
    </id>
    <many-to-one name="address" 
        column="addressId" 
        unique="true"
        not-null="true"/>
</class>

<class name="Address">
    <id name="id" column="addressId">
        <generator class="native"/>
    </id>
</class>

create table Person ( personId bigint not null primary key, addressId bigint not null unique )
create table Address ( addressId bigint not null primary key )
        

A unidirectional one-to-one association on a primary key usually uses a special id generator. (Notice that we've reversed the direction of the association in this example.)

<class name="Person">
    <id name="id" column="personId">
        <generator class="native"/>
    </id>
</class>

<class name="Address">
    <id name="id" column="personId">
        <generator class="foreign">
            <param name="property">person</param>
        </generator>
    </id>
    <one-to-one name="person" constrained="true"/>
</class>

create table Person ( personId bigint not null primary key )
create table Address ( personId bigint not null primary key )
        

A unidirectional one-to-many association on a foreign key is a very unusual case, and is not really recommended.

<class name="Person">
    <id name="id" column="personId">
        <generator class="native"/>
    </id>
    <set name="addresses">
        <key column="personId" 
            not-null="true"/>
        <one-to-many class="Address"/>
    </set>
</class>

<class name="Address">
    <id name="id" column="addressId">
        <generator class="native"/>
    </id>
</class>

create table Person ( personId bigint not null primary key )
create table Address ( addressId bigint not null primary key, personId bigint not null )
        

We think it's better to use a join table for this kind of association.

A unidirectional one-to-many association on a join table is much preferred. Notice that by specifying unique="true", we have changed the multiplicity from many-to-many to one-to-many.

<class name="Person">
    <id name="id" column="personId">
        <generator class="native"/>
    </id>
    <set name="addresses" table="PersonAddress">
        <key column="personId"/>
        <many-to-many column="addressId"
            unique="true"
            class="Address"/>
    </set>
</class>

<class name="Address">
    <id name="id" column="addressId">
        <generator class="native"/>
    </id>
</class>

create table Person ( personId bigint not null primary key )
create table PersonAddress ( personId not null, addressId bigint not null primary key )
create table Address ( addressId bigint not null primary key )
        

A unidirectional many-to-one association on a join table is quite common when the association is optional.

<class name="Person">
    <id name="id" column="personId">
        <generator class="native"/>
    </id>
    <join table="PersonAddress" 
        optional="true">
        <key column="personId" unique="true"/>
        <many-to-one name="address"
            column="addressId" 
            not-null="true"/>
    </join>
</class>

<class name="Address">
    <id name="id" column="addressId">
        <generator class="native"/>
    </id>
</class>

create table Person ( personId bigint not null primary key )
create table PersonAddress ( personId bigint not null primary key, addressId bigint not null )
create table Address ( addressId bigint not null primary key )
        

A unidirectional one-to-one association on a join table is extremely unusual, but possible.

<class name="Person">
    <id name="id" column="personId">
        <generator class="native"/>
    </id>
    <join table="PersonAddress" 
        optional="true">
        <key column="personId" 
            unique="true"/>
        <many-to-one name="address"
            column="addressId" 
            not-null="true"
            unique="true"/>
    </join>
</class>

<class name="Address">
    <id name="id" column="addressId">
        <generator class="native"/>
    </id>
</class>

create table Person ( personId bigint not null primary key )
create table PersonAddress ( personId bigint not null primary key, addressId bigint not null unique )
create table Address ( addressId bigint not null primary key )
        

Finally, we have a unidirectional many-to-many association.

<class name="Person">
    <id name="id" column="personId">
        <generator class="native"/>
    </id>
    <set name="addresses" table="PersonAddress">
        <key column="personId"/>
        <many-to-many column="addressId"
            class="Address"/>
    </set>
</class>

<class name="Address">
    <id name="id" column="addressId">
        <generator class="native"/>
    </id>
</class>

create table Person ( personId bigint not null primary key )
create table PersonAddress ( personId bigint not null, addressId bigint not null, primary key (personId, addressId) )
create table Address ( addressId bigint not null primary key )
        

A bidirectional many-to-one association is the most common kind of association. (This is the standard parent/child relationship.)

<class name="Person">
    <id name="id" column="personId">
        <generator class="native"/>
    </id>
    <many-to-one name="address" 
        column="addressId"
        not-null="true"/>
</class>

<class name="Address">
    <id name="id" column="addressId">
        <generator class="native"/>
    </id>
    <set name="people" inverse="true">
        <key column="addressId"/>
        <one-to-many class="Person"/>
    </set>
</class>

create table Person ( personId bigint not null primary key, addressId bigint not null )
create table Address ( addressId bigint not null primary key )
        

If you use a List (or other indexed collection) you need to set the key column of the foreign key to not null, and let Hibernate manage the association from the collections side to maintain the index of each element (making the other side virtually inverse by setting update="false" and insert="false"):

<class name="Person">
   <id name="id"/>
   ...
   <many-to-one name="address"
      column="addressId"
      not-null="true"
      insert="false"
      update="false"/>
</class>

<class name="Address">
   <id name="id"/>
   ...
   <list name="people">
      <key column="addressId" not-null="true"/>
      <list-index column="peopleIdx"/>
      <one-to-many class="Person"/>
   </list>
</class>

It is important that you define not-null="true" on the <key> element of the collection mapping if the underlying foreign key column is NOT NULL. Don't only declare not-null="true" on a possible nested <column> element, but on the <key> element.

A bidirectional one-to-one association on a foreign key is quite common.

<class name="Person">
    <id name="id" column="personId">
        <generator class="native"/>
    </id>
    <many-to-one name="address" 
        column="addressId" 
        unique="true"
        not-null="true"/>
</class>

<class name="Address">
    <id name="id" column="addressId">
        <generator class="native"/>
    </id>
   <one-to-one name="person" 
        property-ref="address"/>
</class>

create table Person ( personId bigint not null primary key, addressId bigint not null unique )
create table Address ( addressId bigint not null primary key )
        

A bidirectional one-to-one association on a primary key uses the special id generator.

<class name="Person">
    <id name="id" column="personId">
        <generator class="native"/>
    </id>
    <one-to-one name="address"/>
</class>

<class name="Address">
    <id name="id" column="personId">
        <generator class="foreign">
            <param name="property">person</param>
        </generator>
    </id>
    <one-to-one name="person" 
        constrained="true"/>
</class>

create table Person ( personId bigint not null primary key )
create table Address ( personId bigint not null primary key )
        

A bidirectional one-to-many association on a join table. Note that the inverse="true" can go on either end of the association, on the collection, or on the join.

<class name="Person">
    <id name="id" column="personId">
        <generator class="native"/>
    </id>
    <set name="addresses" 
        table="PersonAddress">
        <key column="personId"/>
        <many-to-many column="addressId"
            unique="true"
            class="Address"/>
    </set>
</class>

<class name="Address">
    <id name="id" column="addressId">
        <generator class="native"/>
    </id>
    <join table="PersonAddress" 
        inverse="true" 
        optional="true">
        <key column="addressId"/>
        <many-to-one name="person"
            column="personId"
            not-null="true"/>
    </join>
</class>

create table Person ( personId bigint not null primary key )
create table PersonAddress ( personId bigint not null, addressId bigint not null primary key )
create table Address ( addressId bigint not null primary key )
        

A bidirectional one-to-one association on a join table is extremely unusual, but possible.

<class name="Person">
    <id name="id" column="personId">
        <generator class="native"/>
    </id>
    <join table="PersonAddress" 
        optional="true">
        <key column="personId" 
            unique="true"/>
        <many-to-one name="address"
            column="addressId" 
            not-null="true"
            unique="true"/>
    </join>
</class>

<class name="Address">
    <id name="id" column="addressId">
        <generator class="native"/>
    </id>
    <join table="PersonAddress" 
        optional="true"
        inverse="true">
        <key column="addressId" 
            unique="true"/>
        <many-to-one name="person"
            column="personId" 
            not-null="true"
            unique="true"/>
    </join>
</class>

create table Person ( personId bigint not null primary key )
create table PersonAddress ( personId bigint not null primary key, addressId bigint not null unique )
create table Address ( addressId bigint not null primary key )
        

Finally, we have a bidirectional many-to-many association.

<class name="Person">
    <id name="id" column="personId">
        <generator class="native"/>
    </id>
    <set name="addresses" table="PersonAddress">
        <key column="personId"/>
        <many-to-many column="addressId"
            class="Address"/>
    </set>
</class>

<class name="Address">
    <id name="id" column="addressId">
        <generator class="native"/>
    </id>
    <set name="people" inverse="true" table="PersonAddress">
        <key column="addressId"/>
        <many-to-many column="personId"
            class="Person"/>
    </set>
</class>

create table Person ( personId bigint not null primary key )
create table PersonAddress ( personId bigint not null, addressId bigint not null, primary key (personId, addressId) )
create table Address ( addressId bigint not null primary key )
        

Suppose we have an interface Payment, with implementors CreditCardPayment, CashPayment, ChequePayment. The table per hierarchy mapping would look like:

<class name="Payment" table="PAYMENT">
    <id name="id" type="long" column="PAYMENT_ID">
        <generator class="native"/>
    </id>
    <discriminator column="PAYMENT_TYPE" type="string"/>
    <property name="amount" column="AMOUNT"/>
    ...
    <subclass name="CreditCardPayment" discriminator-value="CREDIT">
        <property name="creditCardType" column="CCTYPE"/>
        ...
    </subclass>
    <subclass name="CashPayment" discriminator-value="CASH">
        ...
    </subclass>
    <subclass name="ChequePayment" discriminator-value="CHEQUE">
        ...
    </subclass>
</class>

Exactly one table is required. There is one big limitation of this mapping strategy: columns declared by the subclasses, such as CCTYPE, may not have NOT NULL constraints.

A table per subclass mapping would look like:

<class name="Payment" table="PAYMENT">
    <id name="id" type="long" column="PAYMENT_ID">
        <generator class="native"/>
    </id>
    <property name="amount" column="AMOUNT"/>
    ...
    <joined-subclass name="CreditCardPayment" table="CREDIT_PAYMENT">
        <key column="PAYMENT_ID"/>
        <property name="creditCardType" column="CCTYPE"/>
        ...
    </joined-subclass>
    <joined-subclass name="CashPayment" table="CASH_PAYMENT">
        <key column="PAYMENT_ID"/>
        ...
    </joined-subclass>
    <joined-subclass name="ChequePayment" table="CHEQUE_PAYMENT">
        <key column="PAYMENT_ID"/>
        ...
    </joined-subclass>
</class>

Four tables are required. The three subclass tables have primary key associations to the superclass table (so the relational model is actually a one-to-one association).

Note that Hibernate's implementation of table per subclass requires no discriminator column. Other object/relational mappers use a different implementation of table per subclass which requires a type discriminator column in the superclass table. The approach taken by Hibernate is much more difficult to implement but arguably more correct from a relational point of view. If you would like to use a discriminator column with the table per subclass strategy, you may combine the use of <subclass> and <join>, as follow:

<class name="Payment" table="PAYMENT">
    <id name="id" type="long" column="PAYMENT_ID">
        <generator class="native"/>
    </id>
    <discriminator column="PAYMENT_TYPE" type="string"/>
    <property name="amount" column="AMOUNT"/>
    ...
    <subclass name="CreditCardPayment" discriminator-value="CREDIT">
        <join table="CREDIT_PAYMENT">
            <key column="PAYMENT_ID"/>
            <property name="creditCardType" column="CCTYPE"/>
            ...
        </join>
    </subclass>
    <subclass name="CashPayment" discriminator-value="CASH">
        <join table="CASH_PAYMENT">
            <key column="PAYMENT_ID"/>
            ...
        </join>
    </subclass>
    <subclass name="ChequePayment" discriminator-value="CHEQUE">
        <join table="CHEQUE_PAYMENT" fetch="select">
            <key column="PAYMENT_ID"/>
            ...
        </join>
    </subclass>
</class>

The optional fetch="select" declaration tells Hibernate not to fetch the ChequePayment subclass data using an outer join when querying the superclass.

You may even mix the table per hierarchy and table per subclass strategies using this approach:

<class name="Payment" table="PAYMENT">
    <id name="id" type="long" column="PAYMENT_ID">
        <generator class="native"/>
    </id>
    <discriminator column="PAYMENT_TYPE" type="string"/>
    <property name="amount" column="AMOUNT"/>
    ...
    <subclass name="CreditCardPayment" discriminator-value="CREDIT">
        <join table="CREDIT_PAYMENT">
            <property name="creditCardType" column="CCTYPE"/>
            ...
        </join>
    </subclass>
    <subclass name="CashPayment" discriminator-value="CASH">
        ...
    </subclass>
    <subclass name="ChequePayment" discriminator-value="CHEQUE">
        ...
    </subclass>
</class>

For any of these mapping strategies, a polymorphic association to the root Payment class is mapped using <many-to-one>.

<many-to-one name="payment" column="PAYMENT_ID" class="Payment"/>

There are two ways we could go about mapping the table per concrete class strategy. The first is to use <union-subclass>.

<class name="Payment">
    <id name="id" type="long" column="PAYMENT_ID">
        <generator class="sequence"/>
    </id>
    <property name="amount" column="AMOUNT"/>
    ...
    <union-subclass name="CreditCardPayment" table="CREDIT_PAYMENT">
        <property name="creditCardType" column="CCTYPE"/>
        ...
    </union-subclass>
    <union-subclass name="CashPayment" table="CASH_PAYMENT">
        ...
    </union-subclass>
    <union-subclass name="ChequePayment" table="CHEQUE_PAYMENT">
        ...
    </union-subclass>
</class>

Three tables are involved for the subclasses. Each table defines columns for all properties of the class, including inherited properties.

The limitation of this approach is that if a property is mapped on the superclass, the column name must be the same on all subclass tables. (We might relax this in a future release of Hibernate.) The identity generator strategy is not allowed in union subclass inheritance, indeed the primary key seed has to be shared accross all unioned subclasses of a hierarchy.

If your superclass is abstract, map it with abstract="true". Of course, if it is not abstract, an additional table (defaults to PAYMENT in the example above) is needed to hold instances of the superclass.

An alternative approach is to make use of implicit polymorphism:

<class name="CreditCardPayment" table="CREDIT_PAYMENT">
    <id name="id" type="long" column="CREDIT_PAYMENT_ID">
        <generator class="native"/>
    </id>
    <property name="amount" column="CREDIT_AMOUNT"/>
    ...
</class>

<class name="CashPayment" table="CASH_PAYMENT">
    <id name="id" type="long" column="CASH_PAYMENT_ID">
        <generator class="native"/>
    </id>
    <property name="amount" column="CASH_AMOUNT"/>
    ...
</class>

<class name="ChequePayment" table="CHEQUE_PAYMENT">
    <id name="id" type="long" column="CHEQUE_PAYMENT_ID">
        <generator class="native"/>
    </id>
    <property name="amount" column="CHEQUE_AMOUNT"/>
    ...
</class>

Notice that nowhere do we mention the Payment interface explicitly. Also notice that properties of Payment are mapped in each of the subclasses. If you want to avoid duplication, consider using XML entities (e.g. [ <!ENTITY allproperties SYSTEM "allproperties.xml"> ] in the DOCTYPE declartion and &allproperties; in the mapping).

The disadvantage of this approach is that Hibernate does not generate SQL UNIONs when performing polymorphic queries.

For this mapping strategy, a polymorphic association to Payment is usually mapped using <any>.

<any name="payment" meta-type="string" id-type="long">
    <meta-value value="CREDIT" class="CreditCardPayment"/>
    <meta-value value="CASH" class="CashPayment"/>
    <meta-value value="CHEQUE" class="ChequePayment"/>
    <column name="PAYMENT_CLASS"/>
    <column name="PAYMENT_ID"/>
</any>

There is one further thing to notice about this mapping. Since the subclasses are each mapped in their own <class> element (and since Payment is just an interface), each of the subclasses could easily be part of another inheritance hierarchy! (And you can still use polymorphic queries against the Payment interface.)

<class name="CreditCardPayment" table="CREDIT_PAYMENT">
    <id name="id" type="long" column="CREDIT_PAYMENT_ID">
        <generator class="native"/>
    </id>
    <discriminator column="CREDIT_CARD" type="string"/>
    <property name="amount" column="CREDIT_AMOUNT"/>
    ...
    <subclass name="MasterCardPayment" discriminator-value="MDC"/>
    <subclass name="VisaPayment" discriminator-value="VISA"/>
</class>

<class name="NonelectronicTransaction" table="NONELECTRONIC_TXN">
    <id name="id" type="long" column="TXN_ID">
        <generator class="native"/>
    </id>
    ...
    <joined-subclass name="CashPayment" table="CASH_PAYMENT">
        <key column="PAYMENT_ID"/>
        <property name="amount" column="CASH_AMOUNT"/>
        ...
    </joined-subclass>
    <joined-subclass name="ChequePayment" table="CHEQUE_PAYMENT">
        <key column="PAYMENT_ID"/>
        <property name="amount" column="CHEQUE_AMOUNT"/>
        ...
    </joined-subclass>
</class>

Once again, we don't mention Payment explicitly. If we execute a query against the Payment interface - for example, from Payment - Hibernate automatically returns instances of CreditCardPayment (and its subclasses, since they also implement Payment), CashPayment and ChequePayment but not instances of NonelectronicTransaction.

HQL and native SQL queries are represented with an instance of org.hibernate.Query. This interface offers methods for parameter binding, result set handling, and for the execution of the actual query. You always obtain a Query using the current Session:

List cats = session.createQuery(
    "from Cat as cat where cat.birthdate < ?")
    .setDate(0, date)
    .list();

List mothers = session.createQuery(
    "select mother from Cat as cat join cat.mother as mother where cat.name = ?")
    .setString(0, name)
    .list();

List kittens = session.createQuery(
    "from Cat as cat where cat.mother = ?")
    .setEntity(0, pk)
    .list();

Cat mother = (Cat) session.createQuery(
    "select cat.mother from Cat as cat where cat = ?")
    .setEntity(0, izi)
    .uniqueResult();]]

Query mothersWithKittens = (Cat) session.createQuery(
    "select mother from Cat as mother left join fetch mother.kittens");
Set uniqueMothers = new HashSet(mothersWithKittens.list());

A query is usually executed by invoking list(), the result of the query will be loaded completely into a collection in memory. Entity instances retrieved by a query are in persistent state. The uniqueResult() method offers a shortcut if you know your query will only return a single object. Note that queries that make use of eager fetching of collections usually return duplicates of the root objects (but with their collections initialized). You can filter these duplicates simply through a Set.

// fetch ids
Iterator iter = sess.createQuery("from eg.Qux q order by q.likeliness").iterate();
while ( iter.hasNext() ) {
    Qux qux = (Qux) iter.next();  // fetch the object
    // something we couldnt express in the query
    if ( qux.calculateComplicatedAlgorithm() ) {
        // delete the current instance
        iter.remove();
        // dont need to process the rest
        break;
    }
}

Iterator kittensAndMothers = sess.createQuery(
            "select kitten, mother from Cat kitten join kitten.mother mother")
            .list()
            .iterator();

while ( kittensAndMothers.hasNext() ) {
    Object[] tuple = (Object[]) kittensAndMothers.next();
    Cat kitten  = tuple[0];
    Cat mother  = tuple[1];
    ....
}

Iterator results = sess.createQuery(
        "select cat.color, min(cat.birthdate), count(cat) from Cat cat " +
        "group by cat.color")
        .list()
        .iterator();

while ( results.hasNext() ) {
    Object[] row = (Object[]) results.next();
    Color type = (Color) row[0];
    Date oldest = (Date) row[1];
    Integer count = (Integer) row[2];
    .....
}

//named parameter (preferred)
Query q = sess.createQuery("from DomesticCat cat where cat.name = :name");
q.setString("name", "Fritz");
Iterator cats = q.iterate();

//positional parameter
Query q = sess.createQuery("from DomesticCat cat where cat.name = ?");
q.setString(0, "Izi");
Iterator cats = q.iterate();

//named parameter list
List names = new ArrayList();
names.add("Izi");
names.add("Fritz");
Query q = sess.createQuery("from DomesticCat cat where cat.name in (:namesList)");
q.setParameterList("namesList", names);
List cats = q.list();

Query q = sess.createQuery("from DomesticCat cat");
q.setFirstResult(20);
q.setMaxResults(10);
List cats = q.list();

Query q = sess.createQuery("select cat.name, cat from DomesticCat cat " +
                            "order by cat.name");
ScrollableResults cats = q.scroll();
if ( cats.first() ) {

    // find the first name on each page of an alphabetical list of cats by name
    firstNamesOfPages = new ArrayList();
    do {
        String name = cats.getString(0);
        firstNamesOfPages.add(name);
    }
    while ( cats.scroll(PAGE_SIZE) );

    // Now get the first page of cats
    pageOfCats = new ArrayList();
    cats.beforeFirst();
    int i=0;
    while( ( PAGE_SIZE > i++ ) && cats.next() ) pageOfCats.add( cats.get(1) );

}
cats.close()

<query name="eg.DomesticCat.by.name.and.minimum.weight"><![CDATA[
    from eg.DomesticCat as cat
        where cat.name = ?
        and cat.weight > ?
] ]></query>

Query q = sess.getNamedQuery("eg.DomesticCat.by.name.and.minimum.weight");
q.setString(0, name);
q.setInt(1, minWeight);
List cats = q.list();

A collection filter is a special type of query that may be applied to a persistent collection or array. The query string may refer to this, meaning the current collection element.

Collection blackKittens = session.createFilter(
    pk.getKittens(), 
    "where this.color = ?")
    .setParameter( Color.BLACK, Hibernate.custom(ColorUserType.class) )
    .list()
);

The returned collection is considered a bag, and it's a copy of the given collection. The original collection is not modified (this is contrary to the implication of the name "filter", but consistent with expected behavior).

Observe that filters do not require a from clause (though they may have one if required). Filters are not limited to returning the collection elements themselves.

Collection blackKittenMates = session.createFilter(
    pk.getKittens(), 
    "select this.mate where this.color = eg.Color.BLACK.intValue")
    .list();

Even an empty filter query is useful, e.g. to load a subset of elements in a huge collection:

Collection tenKittens = session.createFilter(
    mother.getKittens(), "")
    .setFirstResult(0).setMaxResults(10)
    .list();

HQL is extremely powerful but some developers prefer to build queries dynamically, using an object-oriented API, rather than building query strings. Hibernate provides an intuitive Criteria query API for these cases:

Criteria crit = session.createCriteria(Cat.class);
crit.add( Expression.eq( "color", eg.Color.BLACK ) );
crit.setMaxResults(10);
List cats = crit.list();

The Criteria and the associated Example API are discussed in more detail in Chapter 15, Criteria Queries.

You may express a query in SQL, using createSQLQuery() and let Hibernate take care of the mapping from result sets to objects. Note that you may at any time call session.connection() and use the JDBC Connection directly. If you chose to use the Hibernate API, you must enclose SQL aliases in braces:

List cats = session.createSQLQuery(
    "SELECT {cat.*} FROM CAT {cat} WHERE ROWNUM<10",
    "cat",
    Cat.class
).list();

List cats = session.createSQLQuery(
    "SELECT {cat}.ID AS {cat.id}, {cat}.SEX AS {cat.sex}, " +
           "{cat}.MATE AS {cat.mate}, {cat}.SUBCLASS AS {cat.class}, ... " +
    "FROM CAT {cat} WHERE ROWNUM<10",
    "cat",
    Cat.class
).list()

SQL queries may contain named and positional parameters, just like Hibernate queries. More information about native SQL queries in Hibernate can be found in Chapter 16, Native SQL.

First, don't use the session-per-operation antipattern, that is, don't open and close a Session for every simple database call in a single thread! Of course, the same is true for database transactions. Database calls in an application are made using a planned sequence, they are grouped into atomic units of work. (Note that this also means that auto-commit after every single SQL statement is useless in an application, this mode is intended for ad-hoc SQL console work. Hibernate disables, or expects the application server to do so, auto-commit mode immediately.) Database transactions are never optional, all communication with a database has to occur inside a transaction, no matter if you read or write data. As explained, auto-commit behavior for reading data should be avoided, as many small transactions are unlikely to perform better than one clearly defined unit of work. The latter is also much more maintainable and extensible.

The most common pattern in a multi-user client/server application is session-per-request. In this model, a request from the client is send to the server (where the Hibernate persistence layer runs), a new Hibernate Session is opened, and all database operations are executed in this unit of work. Once the work has been completed (and the response for the client has been prepared), the session is flushed and closed. You would also use a single database transaction to serve the clients request, starting and committing it when you open and close the Session. The relationship between the two is one-to-one and this model is a perfect fit for many applications.

The challenge lies in the implementation. Hibernate provides built-in management of the "current session" to simplify this pattern. All you have to do is start a transaction when a server request has to be processed, and end the transaction before the response is send to the client. You can do this in any way you like, common solutions are ServletFilter, AOP interceptor with a pointcut on the service methods, or a proxy/interception container. An EJB container is a standardized way to implement cross-cutting aspects such as transaction demarcation on EJB session beans, declaratively with CMT. If you decide to use programmatic transaction demarcation, prefer the Hibernate Transaction API shown later in this chapter, for ease of use and code portability.

Your application code can access a "current session" to process the request by simply calling sessionFactory.getCurrentSession() anywhere and as often as needed. You will always get a Session scoped to the current database transaction. This has to be configured for either resource-local or JTA environments, see Section 2.5, “Contextual Sessions”.

Sometimes it is convenient to extend the scope of a Session and database transaction until the "view has been rendered". This is especially useful in servlet applications that utilize a separate rendering phase after the request has been processed. Extending the database transaction until view rendering is complete is easy to do if you implement your own interceptor. However, it is not easily doable if you rely on EJBs with container-managed transactions, as a transaction will be completed when an EJB method returns, before rendering of any view can start. See the Hibernate website and forum for tips and examples around this Open Session in View pattern.

The session-per-request pattern is not the only useful concept you can use to design units of work. Many business processes require a whole series of interactions with the user interleaved with database accesses. In web and enterprise applications it is not acceptable for a database transaction to span a user interaction. Consider the following example:

We call this unit of work, from the point of view of the user, a long running conversation (or application transaction). There are many ways how you can implement this in your application.

A first naive implementation might keep the Session and database transaction open during user think time, with locks held in the database to prevent concurrent modification, and to guarantee isolation and atomicity. This is of course an anti-pattern, since lock contention would not allow the application to scale with the number of concurrent users.

Clearly, we have to use several database transactions to implement the converastion. In this case, maintaining isolation of business processes becomes the partial responsibility of the application tier. A single conversation usually spans several database transactions. It will be atomic if only one of these database transactions (the last one) stores the updated data, all others simply read data (e.g. in a wizard-style dialog spanning several request/response cycles). This is easier to implement than it might sound, especially if you use Hibernate's features:

Both session-per-request-with-detached-objects and session-per-conversation have advantages and disadvantages, we discuss them later in this chapter in the context of optimistic concurrency control.

An application may concurrently access the same persistent state in two different Sessions. However, an instance of a persistent class is never shared between two Session instances. Hence there are two different notions of identity:

Then for objects attached to a particular Session (i.e. in the scope of a Session) the two notions are equivalent, and JVM identity for database identity is guaranteed by Hibernate. However, while the application might concurrently access the "same" (persistent identity) business object in two different sessions, the two instances will actually be "different" (JVM identity). Conflicts are resolved using (automatic versioning) at flush/commit time, using an optimistic approach.

This approach leaves Hibernate and the database to worry about concurrency; it also provides the best scalability, since guaranteeing identity in single-threaded units of work only doesn't need expensive locking or other means of synchronization. The application never needs to synchronize on any business object, as long as it sticks to a single thread per Session. Within a Session the application may safely use == to compare objects.

However, an application that uses == outside of a Session, might see unexpected results. This might occur even in some unexpected places, for example, if you put two detached instances into the same Set. Both might have the same database identity (i.e. they represent the same row), but JVM identity is by definition not guaranteed for instances in detached state. The developer has to override the equals() and hashCode() methods in persistent classes and implement his own notion of object equality. There is one caveat: Never use the database identifier to implement equality, use a business key, a combination of unique, usually immutable, attributes. The database identifier will change if a transient object is made persistent. If the transient instance (usually together with detached instances) is held in a Set, changing the hashcode breaks the contract of the Set. Attributes for business keys don't have to be as stable as database primary keys, you only have to guarantee stability as long as the objects are in the same Set. See the Hibernate website for a more thorough discussion of this issue. Also note that this is not a Hibernate issue, but simply how Java object identity and equality has to be implemented.

Never use the anti-patterns session-per-user-session or session-per-application (of course, there are rare exceptions to this rule). Note that some of the following issues might also appear with the recommended patterns, make sure you understand the implications before making a design decision:

If a Hibernate persistence layer runs in a non-managed environment, database connections are usually handled by simple (i.e. non-DataSource) connection pools from which Hibernate obtains connections as needed. The session/transaction handling idiom looks like this:

// Non-managed environment idiom
Session sess = factory.openSession();
Transaction tx = null;
try {
    tx = sess.beginTransaction();

    // do some work
    ...

    tx.commit();
}
catch (RuntimeException e) {
    if (tx != null) tx.rollback();
    throw e; // or display error message
}
finally {
    sess.close();
}

You don't have to flush() the Session explicitly - the call to commit() automatically triggers the synchronization (depending upon the Section 10.10, “Flushing the Session” for the session. A call to close() marks the end of a session. The main implication of close() is that the JDBC connection will be relinquished by the session. This Java code is portable and runs in both non-managed and JTA environments.

A much more flexible solution is Hibernate's built-in "current session" context management, as described earlier:

// Non-managed environment idiom with getCurrentSession()
try {
    factory.getCurrentSession().beginTransaction();

    // do some work
    ...

    factory.getCurrentSession().getTransaction().commit();
}
catch (RuntimeException e) {
    factory.getCurrentSession().getTransaction().rollback();
    throw e; // or display error message
}

You will very likely never see these code snippets in a regular application; fatal (system) exceptions should always be caught at the "top". In other words, the code that executes Hibernate calls (in the persistence layer) and the code that handles RuntimeException (and usually can only clean up and exit) are in different layers. The current context management by Hibernate can significantly simplify this design, as all you need is access to a SessionFactory. Exception handling is discussed later in this chapter.

Note that you should select org.hibernate.transaction.JDBCTransactionFactory (which is the default), and for the second example "thread" as your hibernate.current_session_context_class.

If your persistence layer runs in an application server (e.g. behind EJB session beans), every datasource connection obtained by Hibernate will automatically be part of the global JTA transaction. You can also install a standalone JTA implementation and use it without EJB. Hibernate offers two strategies for JTA integration.

If you use bean-managed transactions (BMT) Hibernate will tell the application server to start and end a BMT transaction if you use the Transaction API. So, the transaction management code is identical to the non-managed environment.

// BMT idiom
Session sess = factory.openSession();
Transaction tx = null;
try {
    tx = sess.beginTransaction();

    // do some work
    ...

    tx.commit();
}
catch (RuntimeException e) {
    if (tx != null) tx.rollback();
    throw e; // or display error message
}
finally {
    sess.close();
}

If you want to use a transaction-bound Session, that is, the getCurrentSession() functionality for easy context propagation, you will have to use the JTA UserTransaction API directly:

// BMT idiom with getCurrentSession()
try {
    UserTransaction tx = (UserTransaction)new InitialContext()
                            .lookup("java:comp/UserTransaction");

    tx.begin();

    // Do some work on Session bound to transaction
    factory.getCurrentSession().load(...);
    factory.getCurrentSession().persist(...);

    tx.commit();
}
catch (RuntimeException e) {
    tx.rollback();
    throw e; // or display error message
}

With CMT, transaction demarcation is done in session bean deployment descriptors, not programatically, hence, the code is reduced to:

// CMT idiom
 Session sess = factory.getCurrentSession();

 // do some work
 ...

In a CMT/EJB even rollback happens automatically, since an unhandled RuntimeException thrown by a session bean method tells the container to set the global transaction to rollback. This means you do not need to use the Hibernate Transaction API at all with BMT or CMT, and you get automatic propagation of the "current" Session bound to the transaction.

Note that you should choose org.hibernate.transaction.JTATransactionFactory if you use JTA directly (BMT), and org.hibernate.transaction.CMTTransactionFactory in a CMT session bean, when you configure Hibernate's transaction factory. Remember to also set hibernate.transaction.manager_lookup_class. Furthermore, make sure that your hibernate.current_session_context_class is either unset (backwards compatiblity), or set to "jta".

The getCurrentSession() operation has one downside in a JTA environment. There is one caveat to the use of after_statement connection release mode, which is then used by default. Due to a silly limitation of the JTA spec, it is not possible for Hibernate to automatically clean up any unclosed ScrollableResults or Iterator instances returned by scroll() or iterate(). You must release the underlying database cursor by calling ScrollableResults.close() or Hibernate.close(Iterator) explicity from a finally block. (Of course, most applications can easily avoid using scroll() or iterate() at all from the JTA or CMT code.)

If the Session throws an exception (including any SQLException), you should immediately rollback the database transaction, call Session.close() and discard the Session instance. Certain methods of Session will not leave the session in a consistent state. No exception thrown by Hibernate can be treated as recoverable. Ensure that the Session will be closed by calling close() in a finally block.

The HibernateException, which wraps most of the errors that can occur in a Hibernate persistence layer, is an unchecked exception (it wasn't in older versions of Hibernate). In our opinion, we shouldn't force the application developer to catch an unrecoverable exception at a low layer. In most systems, unchecked and fatal exceptions are handled in one of the first frames of the method call stack (i.e. in higher layers) and an error message is presented to the application user (or some other appropriate action is taken). Note that Hibernate might also throw other unchecked exceptions which are not a HibernateException. These are, again, not recoverable and appropriate action should be taken.

Hibernate wraps SQLExceptions thrown while interacting with the database in a JDBCException. In fact, Hibernate will attempt to convert the eexception into a more meningful subclass of JDBCException. The underlying SQLException is always available via JDBCException.getCause(). Hibernate converts the SQLException into an appropriate JDBCException subclass using the SQLExceptionConverter attached to the SessionFactory. By default, the SQLExceptionConverter is defined by the configured dialect; however, it is also possible to plug in a custom implementation (see the javadocs for the SQLExceptionConverterFactory class for details). The standard JDBCException subtypes are:

One extremely important feature provided by a managed environment like EJB that is never provided for non-managed code is transaction timeout. Transaction timeouts ensure that no misbehaving transaction can indefinitely tie up resources while returning no response to the user. Outside a managed (JTA) environment, Hibernate cannot fully provide this functionality. However, Hibernate can at least control data access operations, ensuring that database level deadlocks and queries with huge result sets are limited by a defined timeout. In a managed environment, Hibernate can delegate transaction timeout to JTA. This functioanlity is abstracted by the Hibernate Transaction object.

Session sess = factory.openSession();
try {
    //set transaction timeout to 3 seconds
    sess.getTransaction().setTimeout(3);
    sess.getTransaction().begin();

    // do some work
    ...

    sess.getTransaction().commit()
}
catch (RuntimeException e) {
    sess.getTransaction().rollback();
    throw e; // or display error message
}
finally {
    sess.close();
}

Note that setTimeout() may not be called in a CMT bean, where transaction timeouts must be defined declaratively.

In an implementation without much help from Hibernate, each interaction with the database occurs in a new Session and the developer is responsible for reloading all persistent instances from the database before manipulating them. This approach forces the application to carry out its own version checking to ensure conversation transaction isolation. This approach is the least efficient in terms of database access. It is the approach most similar to entity EJBs.

// foo is an instance loaded by a previous Session
session = factory.openSession();
Transaction t = session.beginTransaction();

int oldVersion = foo.getVersion();
session.load( foo, foo.getKey() ); // load the current state
if ( oldVersion!=foo.getVersion ) throw new StaleObjectStateException();
foo.setProperty("bar");

t.commit();
session.close();

The version property is mapped using <version>, and Hibernate will automatically increment it during flush if the entity is dirty.

Of course, if you are operating in a low-data-concurrency environment and don't require version checking, you may use this approach and just skip the version check. In that case, last commit wins will be the default strategy for your long conversations. Keep in mind that this might confuse the users of the application, as they might experience lost updates without error messages or a chance to merge conflicting changes.

Clearly, manual version checking is only feasible in very trivial circumstances and not practical for most applications. Often not only single instances, but complete graphs of modified ojects have to be checked. Hibernate offers automatic version checking with either an extended Session or detached instances as the design paradigm.

A single Session instance and its persistent instances are used for the whole conversation, known as session-per-conversation. Hibernate checks instance versions at flush time, throwing an exception if concurrent modification is detected. It's up to the developer to catch and handle this exception (common options are the opportunity for the user to merge changes or to restart the business conversation with non-stale data).

The Session is disconnected from any underlying JDBC connection when waiting for user interaction. This approach is the most efficient in terms of database access. The application need not concern itself with version checking or with reattaching detached instances, nor does it have to reload instances in every database transaction.

// foo is an instance loaded earlier by the old session
Transaction t = session.beginTransaction(); // Obtain a new JDBC connection, start transaction

foo.setProperty("bar");

session.flush();    // Only for last transaction in conversation
t.commit();         // Also return JDBC connection
session.close();    // Only for last transaction in conversation

The foo object still knows which Session it was loaded in. Beginning a new database transaction on an old session obtains a new connection and resumes the session. Committing a database transaction disconnects a session from the JDBC connection and returns the connection to the pool. After reconnection, to force a version check on data you aren't updating, you may call Session.lock() with LockMode.READ on any objects that might have been updated by another transaction. You don't need to lock any data that you are updating. Usually you would set FlushMode.NEVER on an extended Session, so that only the last database transaction cycle is allowed to actually persist all modifications made in this conversation. Hence, only this last database transaction would include the flush() operation, and then also close() the session to end the conversation.

This pattern is problematic if the Session is too big to be stored during user think time, e.g. an HttpSession should be kept as small as possible. As the Session is also the (mandatory) first-level cache and contains all loaded objects, we can probably use this strategy only for a few request/response cycles. You should use a Session only for a single conversation, as it will soon also have stale data.

(Note that earlier Hibernate versions required explicit disconnection and reconnection of a Session. These methods are deprecated, as beginning and ending a transaction has the same effect.)

Also note that you should keep the disconnected Session close to the persistence layer. In other words, use an EJB stateful session bean to hold the Session in a three-tier environment, and don't transfer it to the web layer (or even serialize it to a separate tier) to store it in the HttpSession.

The extended session pattern, or session-per-conversation, is more difficult to implement with automatic current session context management. You need to supply your own implementation of the CurrentSessionContext for this, see the Hibernate Wiki for examples.

Each interaction with the persistent store occurs in a new Session. However, the same persistent instances are reused for each interaction with the database. The application manipulates the state of detached instances originally loaded in another Session and then reattaches them using Session.update(), Session.saveOrUpdate(), or Session.merge().

// foo is an instance loaded by a previous Session
foo.setProperty("bar");
session = factory.openSession();
Transaction t = session.beginTransaction();
session.saveOrUpdate(foo); // Use merge() if "foo" might have been loaded already
t.commit();
session.close();

Again, Hibernate will check instance versions during flush, throwing an exception if conflicting updates occured.

You may also call lock() instead of update() and use LockMode.READ (performing a version check, bypassing all caches) if you are sure that the object has not been modified.

You may disable Hibernate's automatic version increment for particular properties and collections by setting the optimistic-lock mapping attribute to false. Hibernate will then no longer increment versions if the property is dirty.

Legacy database schemas are often static and can't be modified. Or, other applications might also access the same database and don't know how to handle version numbers or even timestamps. In both cases, versioning can't rely on a particular column in a table. To force a version check without a version or timestamp property mapping, with a comparison of the state of all fields in a row, turn on optimistic-lock="all" in the <class> mapping. Note that this concepetually only works if Hibernate can compare the old and new state, i.e. if you use a single long Session and not session-per-request-with-detached-objects.

Sometimes concurrent modification can be permitted as long as the changes that have been made don't overlap. If you set optimistic-lock="dirty" when mapping the <class>, Hibernate will only compare dirty fields during flush.

In both cases, with dedicated version/timestamp columns or with full/dirty field comparison, Hibernate uses a single UPDATE statement (with an appropriate WHERE clause) per entity to execute the version check and update the information. If you use transitive persistence to cascade reattachment to associated entities, Hibernate might execute uneccessary updates. This is usually not a problem, but on update triggers in the database might be executed even when no changes have been made to detached instances. You can customize this behavior by setting select-before-update="true" in the <class> mapping, forcing Hibernate to SELECT the instance to ensure that changes did actually occur, before updating the row.