Red Hat

In Relation To

The Hibernate team blog on everything data.

JDK 1.5 breaks my ObjectFactory

Posted by    |       |    Tagged as

This is /just great/ ...

JDK 1.5 changes the method signature of javax.naming.spi.ObjectFactory.getObjectInstance() from this:

public Object getObjectInstance(Object reference, Name name, Context ctx, Hashtable env)

to this:

public Object getObjectInstance(Object reference, Name name, Context ctx, Hashtable<String, ?> env)

AFAICT, this means that is now impossible to write an ObjectFactory that compiles in both JDK 1.4 and 1.5. Ugh.

Internationalized data in Hibernate

Posted by    |       |    Tagged as

We've seen a few people using internationalized reference data where labels displayed in the user interface depend upon the user's language. It's not immediately obvious how to deal with this in Hibernate, and I've been meaning to write up my preferred solution for a while now.

Suppose I have a table which defines labels in terms of a unique code, together with a language.

create table Label (
    code bigint not null,
    language char(2) not null,
    description varchar(100) not null,
    primary key(code, langauge)

Other entities refer to labels by their code. For example, the Category table needs category descriptions.

create table Category (
    category_id bigint not null primary key,
    discription_code bigint not null,
    parent_category_id foreign key references(category)

Note that for each description_code, there are potentially many matching rows in the Label table. At runtime, my Java Category instances should be loaded with the correct description for the user's language preference.

UI Labels should certainly be cached between transactions. We could implement this cache either in our application, or by mapping a Label class and using Hibernate's second-level cache. How we implement this is not very relevant, we'll assume that we have some cache, and can retrieve a description using:

Label.getDescription(code, language)

And get the code back using:

Label.getCode(description, language)

Our Category class looks like this:

public class Category {
    private Long id;
    private String description;
    private Category parent;

The description field holds the String-valued description of the Category in the user's language. But in the database table, all we have is the code of the description. It seems like this situation can't be handled in the a Hibernate mapping.

Whenever it seems like you can't do something in Hibernate, you should think UserType! We'll use a UserType to solve this problem.

public class LabelUserType {
    public int[] sqlTypes() { return Types.BIGINT; }
    public Class returnedClass() { return String.class; }
    public boolean equals(Object x, Object y) throws HibernateException {
        return x==null ? y==null : x.equals(y);
    public Object nullSafeGet(ResultSet rs, String[] names, Object owner) 
        throws HibernateException, SQLException {
        Long code = (Long) Hibernate.LONG.nullSafeGet(rs, names, owner);
        return Label.getDescrption( code, User.current().getLanguage() );
    public void nullSafeSet(PreparedStatement st, Object value, int index) 
        throws HibernateException, SQLException {
        Long code = Label.getCode( (String) value, User.current().getLanguage() );
        Hibernate.LONG.nullSafeSet(st, code, index);
    public Object deepCopy(Object value) throws HibernateException {
        return value; //strings are immutable
    public boolean isMutable() {
        return false;

(We can get the current user's language preference by calling User.current().getLanguage().)

Now we can map the Category class:

<class name="Categoy">
    <id name="id" column="category_id">
        <generator class="native"/>

Note that we can even write queries against Category.description. For example:

String description = ...;
session.createQuery("from Category c where c.description = :description")
    .setParameter("description", description, Hibernate.custom(LabelUserType.class))

or, to specify the code:

Long code = ...;
session.createQuery("from Category c where c.description = :code")
    .setLong("description", code)

Unfortunately, we can't perform text-based searching using like, nor can we order by the textual description. We would need to perform sorting of (or by) labels in memory.

Notice that this implementation is very efficient, we never need to join to the Label table in our queries - we never need to query that table at all, except at startup time to initialize the cache. A potential problem is keeping the cache up to date if the Label data changes. If you use Hibernate to implement the Label cache, there's no problem. If you implement it in your application, you will need to manually refresh the cache when data changes.

This pattern can be used for more than internationalization, by the way!

Comparing ORM tools

Posted by    |       |    Tagged as Hibernate ORM

I've seen three or four ORM tool comparisons in the last three weeks; on some weblogs, on our forum and I've even been part in several decisions.

I have the impression that many developers have problems categorizing and evaluating ORM tools, no matter if its Hibernate, Cayenne, PrIdE (I hope that spelling is correct), or some home-made JDBC framework. I got really frustrated at some point, but what brings me to this blog entry is probably a posting made today, by Scott Ferguson. He compares EJB CMP, JDO, and Hibernate. I wasn't really happy with his list of points. Don't get me wrong, I'm not complaining about Scott's conclusions (our precious Hibernate!), in fact, I actually usually listen to Scott. I've even followed Resins development closely several years ago, nearly got it approved for a medium-sized installation (politics...), and even reported and fixed some bugs.

So, this entry, after a long introduction, is about comparing ORM solutions. What all the reviews and articles had in common was a very very obscure criteria schema. In one article, I've seen someone comparing loading and saving a single object and looking at the lines of code that you need for this operation. Next, we hear something like my ORM should work with objects or other vague statements that, in practice, probably not help you decide what you should use.

I did my research for Hibernate in Action, and I think we have found an excellent taxonomy for ORM solutions. Actually, Mark Fussel started to use these categories in 1997, we merely rewrote his list and set it in context to Java application development:

Pure relational

The whole application, including the user interface, is designed around the relational model and SQL-based relational operations. Direct SQL can be fine-tuned in every aspect, but the drawbacks, such as difficult maintenance, lack of portability, and maintainability, are significant, especially in the long run. Applications in this category often make heavy use of stored procedures, shifting some of the work out of the business layer and into the database.

Light object mapping

Entities are represented as classes that are mapped manually to the relational tables. Hand-coded SQL/JDBC is hidden from the business logic using well-known design patterns (such as DAO). This approach is extremely widespread and is successful for applications with a small number of entities, or applications with generic, metadata-driven data models. Stored procedures might have a place in this kind of application.

Medium object mapping

The application is designed around an object model. SQL is generated at build time using a code generation tool, or at runtime by framework code. Associations between objects are supported by the persistence mechanism, and queries may be specified using an object-oriented expression language. Objects are cached by the persistence layer. A great many ORM products and homegrown persistence layers support at least this level of functionality. It's well suited to medium-sized applications with some complex transactions, particularly when portability between different database products is important. These applications usually don't use stored procedures.

Full object mapping

Full object mapping supports sophisticated object modeling: composition, inheritance, polymorphism, and persistence by reachability or a more flexible transitive persistence solution. The persistence layer implements transparent persistence; persistent classes do not inherit any special base class or have to implement a special interface. The persistence layer does not enforce a particular programming model for the domain model implementation. Efficient fetching strategies (lazy and eager fetching) and caching strategies are implemented transparently to the application. This level of functionality can hardly be achieved by a homegrown persistence layer - it's equivalent to months or years of development time.

In my experience, it is quite easy to find the category for a given product. In Hibernate in Action, we also have a list of interesting questions that you should ask if you compare ORM tools:

  • What do persistent classes look like? Are they fine-grained JavaBeans?
  • How is mapping metadata defined?
  • How should we map class inheritance hierarchies?
  • How does the persistence logic interact at runtime with the objects of the business domain?
  • What is the lifecycle of a persistent object?
  • What facilities are provided for sorting, searching, and aggregating?
  • How do we efficiently retrieve data with associations?

In addition, two issues are common to any data-access technology. They also impose fundamental constraints on the design and architecture of an ORM:

  • Transactions and concurrency
  • Cache management (and concurrency)

Find the answers to those questions, and you can compare ORM software. Scott in fact started right with the lifecycle, but he has not given enough information in his article for a real discussion, it's mostly his opinion (which is fine on a weblog).

There are, as always in life, many solutions and not a single product, project, or specification will be perfect in all scenarios. You don't have to try to get to the top of the list and always use Full object mapping (and the appropriate tool). There are very good reasons to use a Light object mapping tool (iBatis, for example) in some situations! In many situations, JDBC and SQL are the best choice. I'm talking about a comparison at the same level, and I've made good experience with the categories and questions I've shown above. Read the book. :)

Thanks for listening

History triggers and Hibernate

Posted by    |       |    Tagged as

Recently, I helped one of our customers migrating a legacy database to Hibernate; one of the more interesting topics was versioning and audit logging. Actually, in the last couple of months, the subject of historical data came up several times. No matter if it was a legacy SQL schema or a migration from a broken object-oriented database, everyone had their own way to log data changes.

In this entry, I'll introduce a clean and nice solution for this issue. My proposal naturally integrates with Hibernate. Let's use database triggers and views instead of code in the application layer.

While it is in fact quite easy to write a Hibernate Interceptor for audit logging (an example can be found in Hibernate in Action or on the Hibernate Wiki ), we always like to use the features of the database system. Implementing audit logging in the database is the best choice if many applications share the same schema and data, and usually much less hassle to maintain in the long run.

First, let's create an entity we want to implement a change history for, a simple Item. In Java, this entity is implemented as the Item class. As usual for a Hibernate application that uses Detached Objects and automatic optimistic concurrency control, we give it an id and a version property:

public class Item {

    private Long id = null
    private int version;
    private String description;
    private BigDecimal price;

    Item() {}
    ... // Accessor and business methods    

This class is then mapped to a table using Hibernate metadata:

<class name="Item" table="ITEM_VERSIONED>
    <id name="id" column="ITEM_ID">
        <generator class="native"/>
    <version name="version" column="VERSION"/>
    <property name="description" column="DESC"/>
    <property name="price" column="PRICE"/>

The name of the mapped table is ITEM_VERSIONED. This is actually not a normal base table, but a database view that joins the data from two base tables. Let's have a look at the two tables in Oracle:

create table ITEM (
    DESC       VARCHAR(255) NOT NULL,
    PRICE      NUMBER(19,2) NOT NULL,

create table ITEM_HISTORY (
    DESC       VARCHAR(255) NOT NULL,
    PRICE      NUMBER(19,2) NOT NULL,

The ITEM table is our real entity relation. The ITEM_HISTORY table has a different primary key, using the ITEM_ID and VERSION column. Our goal is to have one row per entity instance in ITEM (the newest version of our data) and one row for each item version in ITEM_HISTORY:

ITEM_ID   DESC            PRICE
1         A nice Item.    123,99
2         Another one.     34,44

ITEM_ID   DESC            PRICE      VERSION
1         The original.   123,99     0
1         An update.      123,99     1
1         A nice Item.    123,99     2
2         Another one.     34,44     0

So, instead of mapping our Java entity to any of the two tables, we map it to a new virtual table, ITEM_VERSIONED. This view merges the data from both base tables:

create or replace view ITEM_VERSIONED (ITEM_ID, VERSION, DESC, PRICE) as
    select I.ITEM_ID as ITEM_ID,
        (select max(IH.VERSION)
            from ITEM_HISTORY HI
            where HI.ITEM_ID = I.ITEM_ID) as VERSION,
        I.DESC as DESC,
        I.PRICE as PRICE
    from   ITEM I

The ITEM_VERSIONED view uses a correlated subquery and a theta-style join to get the highest version number for a particular item from the history table, while selecting the current values from the row in ITEM. Of course we could also directly read all data from ITEM_HISTORY, but this query is more flexible, for example if you don't want to include all original columns in the history.

Hibernate can now read entities and it has a version number for automatic optimistic locking. However, we can not save entities, since the view is read-only. (In Oracle and most other databases, views created using a join can not be updated.) You will get an exception if you try to update an entity.

We solve this problem by writing a database trigger. The trigger will intercept all updates and insertions for the view and redirect the data to the base tables. This kind of trigger is called an /INSTEAD OF/ trigger. Let's first handle insertion:

create or replace trigger ITEM_INSERT
    instead of insert on ITEM_VERSIONED begin
    insert into ITEM(ITEM_ID, DESC, PRICE)
           values (:n.ITEM_ID, :n.DESC, :n.PRICE);
           values (:n.ITEM_ID, :n.DESC, :n.PRICE, :n.VERSION);

This trigger will execute two inserts and split the data between the entity and entity history table. Next, update operations:

create or replace trigger ITEM_UPDATE
    instead of update on ITEM_VERSIONED begin
    update ITEM set
            DESC = :n.DESC,
            PRICE = :n.PRICE,
            ITEM_ID = :n.ITEM_ID;
           values (:n.ITEM_ID, :n.DESC, :n.PRICE, :n.VERSION);

The entity table is updated first, with the new data. Then, a new row is written to the ITEM_HISTORY table.

This is actually all you need to implement a basic history functionality, just check /INSTEAD OF/ trigger support in your database management system. You can even enhance this pattern and make it much more flexible: write a new Auditinfo value type class with user and timestamp information and add an auditinfo property to your entity class in Java. Map this to some new columns in your view using a Hibernate custom UserType and track the information by setting the property in a Hibernate Interceptor when updates and inserts occur. Use AOP to externalize this aspect from your POJOs...



Posted by    |       |    Tagged as

Abe White of Solarmetric replies to my criticisms of JDO on TSS. I'm actually not interested in getting into a lengthy debate over this, but since there /was/ an error in my first post, I must certainly acknowledge that.

First, Abe displays a trivial query that looks superficially similar in HQL and JDOQL. I'm not certain exactly what this is intended to prove, but I'm not going to get into a lengthy debate over this. I encourage anyone who is interested to compare the two query languages for themselves. It is my firm belief that a query language for ORM should look like SQL. (Unlike HQL and EJBQL, JDOQL is not targetted specifically at ORM, which explains some of the difference of opinion.) I guess I should make it clear that it is really the query language that is the showstopper for me personally.

Next, Abe points out that I am wrong about the semantics for unfetched associations in JDO detachment. I stand corrected. From the JDO1 spec, and from earlier conversations with JDO EG members, I had understood that the JDO guys were insistent that enhanced classes not be needed on the client - which means that interception could not be done for serialized instances on the client side. Apparently they have changed their minds and dropped that requirement. That's reasonable, I suppose. It probably means that there would be difficulties with enhancing at classloading time, but I'm not certain of this and I do accept that build-time enhancement is a reasonable approach to supporting detach/reattach.

There was some suggestion that my other objections to field interception might be wrong, but I think I'm right on those. I think you'll see this if you carefully consider how polymorphic associations and detachment complicate the picture. (Remember, we cannot decide the concrete subclass of an associated instance without hitting its database table or tables.)

Last, Abe argues that JDO really only has exactly the same kinds of identity and instance states as Hibernate. I agree - and that's the whole problem! Let's take identity. There is really just one kind of identity that is interesting: persistent identity. The only thing that should vary is how the identity value (primary key value) is assigned. But JDO has a different representation for datastore identity (a surrogate key generated by the persistence layer) and application identity (a natural key assigned by the application). JDO2 adds simple identity to paper over the problems with this, and complicates the picture further.

OK, no time for more, I gotta catch a plane. Of course, everyone thinks their own technology is best and that everyone who criticizes it is wrong/stupid/spreading FUD/etc. I have absolutely zero expectation that a debate about this will produce any clear technical winner (rather than smeared reputations). I'm fallible, the JDO guys are fallible, everyone else is fallible. I prefer to surrender to darwinism, and watch what solution is actually adopted in real projects.


Posted by    |       |    Tagged as

Yesterday, Linda DeMichiel announced the changes coming in EJB 3.0. There was a lot to digest in her presentation, and I think it will take a while for people to figure out the full implications of the new spec. So far, most attention has focused upon the redesign of entity beans, but that is most certainly not all that is new! The expert group has embraced annotations aggressively, finally eliminating deployment descriptor XML hell. Taking a leaf from Avalon, Pico, Spring, Hivemind, etc, EJB will use dependency injection as an alternative to JNDI lookups. Session beans will be POJOs, with a business interface, home objects have been eliminated. Along with various other changes, this means that EJB 3.0 will be a much more appropriate solution for web-based applications with servlets and business logic colocated in the same process (which is by far the most sane deployment topology for most - but not all - applications), without losing the ability to handle more complex distributed physical architectures.

What is amazing is the broad consensus in the EJB Expert Group - all the way from traditional J2EE vendors like BEA, to we open source kiddies - about what was most needed in EJB3. There has been a lot of listening to users going on, which took me by surprise. Linda's leadership is also to be credited.

But anyway, this is the Hibernate blog, so I'm going to discuss persistence....

EJB 3.0 adopted a POJO-based entity bean programming model, very much similar to what we use in Hibernate. Entity beans may be serializable. They will not need to extend or implement any interfaces from javax.ejb. They will be concrete classes, with JavaBeans-style property accessors. Associations will be of type Set or Collection and always bidirectional, but un-managed. (Referential integrity of bidirectional associations is trivial to implement in your object model, and then applies even when the persistent objects are detached.) This model facilitates test-driven development, allows re-use of the domain model outside the context of the EJB container (especially, DTOs will be a thing of the past for many applications) and emphasizes the business problem, not the container.

Hibernate Query Language was originally based upon EJBQL, ANSI SQL and ODMG OQL, and now the circle is complete, with features of HQL (originally stolen from SQL and OQL) making their way back into EJBQL. These features include explicit joins (including outer joins), projection, aggregation, subselects. No more Fast Lane Reader type antipatterns!

In addition, EJBQL will grow support for bulk update and bulk delete (a feature that we do not currently have). Many users have requested this.

For the occasional cases where the newly enhanced EJBQL is not enough, you will be able to write queries in native SQL, and have the container return managed entities.

EJB 3.0 will replace entity bean homes with a singleton EntityManager object. Entities may be instantiated using new, and then made persistent by calling create(). They may be made transient by calling remove(). The EntityManager is a factory for Query objects, which may execute named queries defined in metadata, or dynamic queries defined via embedded strings or string manipulation. The EntityManager is very similar to a Hibernate Session, JDO PersistenceManager, TopLink UnitOfWork or ODMG Database - so this is a very well-established pattern! Association-level cascade styles will provide for cascade save and delete.

There will be a full ORM metadata specification, defined in terms of annotations. Inheritance will finally be supported, and perhaps some nice things like derived properties.

Because everyone will ask....

What was wrong with JDO? Well, the EJB expert group - before I joined - came to the conclusion that JDO was just not a very appropriate model for ORM, which echoes the conclusion reached by the Hibernate team, and many other people. There are some problems right at the heart of the JDO spec that are simply not easy to fix.

I'm sure I will cop enormous amounts of flak for coming out and talking about these problems in public, but I feel we need to justify this decision to the community, since it affects the community, and since the EG should be answerable to the community. We also need to put to rest the impression that this was just a case of Not Invented Here.

First, JDOQL is an abomination. (There I said it.) There are four standard ways of expressing object-oriented queries: query language, query by criteria, query by example and native SQL. JDOQL is none of these. I have no idea how the JDO EG arrived at the design they chose, but it sort of looks as if they couldn't decide between query language and query by criteria, so went down some strange middle road that exhibits the advantages of neither approach. My suggestion to adopt something like HQL was a nonstarter. The addition of support for projection and aggregation in JDO2 makes JDOQL even uglier and more complex than before. This is /not/ the solution we need!

Second, field interception - which is a great way to implement stuff like Bill Burke's ACID POJOs or the fine-grained cache replication in JBossCache - turns out, perhaps surprisingly, to be a completely inappropriate way to implement POJO persistence. The biggest problem rears its head when we combine lazy association fetching with detached objects. In a proxy-based solution, we throw an exception from unfetched associations if they are accessed outside the context of the container. JDO represents an unfetched association using null. This, at best, means you get a meaningless NPE instead of a LazyInitializationException. At worst, your code might misinterpret the semantics of the null and assume that there is no associated object. This is simply unnacceptable, and there does not appear to be any way to fix JDO to remedy this problem, without basically rearchitecting JDO. (Unlike Hibernate and TopLink, JDO was not designed to support detachment and reattachment.)

Proxy-base solutions have some other wonderful advantages, such as the ability to create an association to an object without actually fetching it from the database, and the ability to discover the primary key of an associated object without fetching it. These are both very useful features.

Finally, the JDO spec is just absurdly over-complex, defining three - now four - types of identity, where one would do, umpteen lifecycle states and transitions, when there should be just three states (persistent, transient, detached) and is bloated with useless features such as transient transactional instances. Again, this stuff is just not easy to change - the whole spec would need to be rewritten.

So, rather than rewriting JDO, EJB 3.0 entities will be based closely upon the (much more widely adopted) dedicated ORM solutions such as Hibernate and TopLink.

Practical tips on JDK 1.5 annotations

Posted by    |       |    Tagged as

Maybe you haven't noticed, but every single piece of information on JDK 1.5 Metadata Annotations is nothing but a simple /Hello World/ tutorial (of course, I exclude the specification). There is no real world implementation. This is especially true for codes processing the annotations. I'll call such an application an annotation reader.

I was publicly cursing about that and Cedric Beust answered in his blog and gave me some hints.

I want to add some more tips and comments about Cedric's list, based on my experience with the development of Hibernate annotations:

  • default values, what about null?
  • faking a non-existing annotation programmatically
  • no extends, what about code duplication?

default values, what about null?

Nothing to declare, except that Cedric is right. The whole point is that there should be something that let's the annotation reader know whether the user set the value or used the default one. However, no such mechanism has been provided by the JSR.

As a workaround I've also build my own BooleanImplied enum type that takes care of values not explicitly set by the user. With Strings, I don't really like using some restricted key-string, such as #implied. For now, my special key-string works like a charm, but one day it will face the what-if-I-want-to-use-that-keyword-as-a-value syndrome. I've no better answer :-(

faking a non existing annotation programmatically

In annotation reader code, there is no way to programmatically build a new annotation /instance/, whether with or without default values. Of course, an annotation is an interface and there is no implementation provided! However, building your own annotation could be useful to keep the code as generic as possible and to fake a user provided annotation. (Especially if most of your annotations have default values).

Let's have a look at an example:

public @interface Column {
     * column name
     * default value means the name is guessed.
    String name() default "#implied";

     * column length
     * Default value 255  
    int length() default 255;


public @interface Property {

     * column(s) used
     * Default is guessed programmatically based on the default @Column()
    Column[] columns() default {};


public String getMyProperty() {
    return myProperty;

I want to use the default values of @Column() when the user does not set any @Column() array in @property(). Why? Because I want the simplest annotations, I want implicit and default values to be the rule, and I don't want to duplicated my annotation processing code.

Two solutions:

  • duplicate the default values somewhere in a value holder class (don't like that one)
  • keep a fake method somewhere with the default @Column() inside:
 * Get default value for annotation here
 * (I couldn't find a way to build an annotation programatically)
private static void defaultColumnAnnHolder() {}

You'll then be able to read and reuse this method in your annotation reader code. I think some bytecode manipulation framework can easily and elegantly provide such feature without the ugly extra method in your code. (Javassist 3.0 should be able to do that if I understand Bill Burke, but I've not tested it yet)

no extends, what about code duplication?

That one that really sucks. The extends keyword is not allowed for annotation. Without it, we can't build a generic method, handling an annotation hierarchy. Bye-bye to the beauty of OO.

Suppose I have @CoreProperty() and then @ComplexProperty():

public void handleProperties(CoreProperty ann) {

This can't process @ComplexProperty(), as there is no way to let @ComplexProperty inherit @CoreProperty.

Some solutions:

  • force your user to set @ComplexProperty() and @CoreProperty() on its annoted method (I don't consider this as an option)
  • use composition
  • copy annotation data into a new class hierarchy (or flatten it)
  • use reflection on annotation

The second one is pretty simple but it may not be appropriate for every model:


If the CoreProperty is an implementation concept, it's pretty ugly to show that in the annotation.

The third solution kind of sucks, but that is the one I'm using now. The annotation reader has to be implemented with two layers:

  • a layer reading and flatteninng the annotation (or putting it in an extra class hierarchy)
  • a process method, handling properties in flat (or hierarchical) representation:
public void processPropertyAnn(Ann property) {
    String name =;
    processFlatProperty(name, null);

 * Process a property
 * @param name property name
 * @param extraValue this extraValue can be set using ComplexProperty
private void processFlatProperty(String name, String extraValue) {

This leads to some extra metadata structure: the good old DTO are evil issue appears in annotation!

The fourth solution will probably be complex and unreadable code, it seems to me that reflection is overkill in this case (I haven't implemented it, however).

Time to conclude

Well, I stayed focused on JSR 175 practical flaws and gave some tips to make life easier. I still strongly believe annotations are good and will give us a new way of thinking metadata in our apps. I don't care which one is best: XDoclet or JSR 175, I care about a standard metadata facility (this topic is mature enough for standardization). I care about a standard metatada desription to easily plug tools on top of frameworks like Hibernate.

This spec will evolve. But for now, don't curse about it, use it for real!

Stateful Session Beans Rock

Posted by    |       |    Tagged as

Like, I suppose, many Java developers, I have so often read about the supposed scalability problems associated with stateful session beans, that I simply accepted that these problems were real, and refused to even consider using stateful beans. I guess this was laziness, but we don't have time to verify everything we read - and I'd never had cause to doubt that what I read was correct.

A few months ago, as Christian and I were nailing down our notion of application transactions in a hotel room in Kohn Kaen, Thailand, it really struck me hard that a stateful bean is the perfect place to keep state associated with the application transaction. (An application transaction is a unit of work /from the point of view of the user/; it spans multiple database/JTA transactions.) For example, you could keep a dedicated Hibernate session for the lifetime of the stateful bean, obviating the need for detaching and reattaching the object graph at every request.

This made me wonder about the cause of the scalability problems that everyone talks about. Once I really started thinking about this, it just didn't add up! A stateful bean without failover is quite equivalent to a HttpSession with server affinity, a design that is commonly understood to scale acceptably well. Similarly, failover for stateful beans should be no more difficult to implement than HttpSession failover.

Indeed, it seemed that the use of stateful beans should actually improve performance, since we would no longer need to wear the cost of serializing state relating to the application transaction to and from either the web tier or database /upon each request/. It seems far more natural to keep the state right there, in the middle tier, along with the business logic, where it belongs.

The conclusion I came to at that time was that the scalability problems must be due to implementation problems in existing appservers.

Well, I was talking about this with a friend who works for one of the other J2EE vendors the other day and she pointed me to this excellent paper:

HTTP Session Object vs Stateful EJB

which really debunks that particular superstition.

Actually, there is a lot of nonsense written about the desirability of so-called stateless architectures. It certainly might be true that a truly stateful design has some nice scalability characteristics. The trouble is that a stateless application can't really /do/ anything of consequence. In a real life application, the state /has to live somewhere/. Serializing it to and from the web tier is just a waste of breath. Serializing it to the database is even worse.

In future, I think I'll find myself using stateful beans all the time.

Lesson: beware J2EE folklore!

The Scope of Hibernate Three

Posted by    |       |    Tagged as Java EE

After more than a year of activity, development of the Hibernate2 branch has finally been wound up; Hibernate 2.1.3 will be one of the last releases and represents a rock-solid POJO persistence solution with essentially all the functionality needed by a typical Java application. Any future release of Hibernate 2.1 will contain only bugfixes. The branch that we have been calling 2.2, will actually be released as version 3.

The Hibernate project has previously had quite circumscribed goals - we have limited ourselves to thinking about just the very next release, and just what our users are asking for right now. That approach has finally reached the end of its usefulness. Hibernate3 is concieved in hubris, with the goal of innovating beyond what our users are asking for, or have even thought of. Certainly we will be adding features that go well beyond the functionality of the best commercial ORM solutions such as TopLink. (Some of these features will be most interesting to people working with certain specialized kinds of problems.)

So, in this spirit of hubris, I'll drop my normal policy of not boasting about things we have not yet implemented, and give a sketch of some of the things that are planned for Hibernate3 alpha (as usual, we do not commit to dates for future releases).


A number of important and interesting problems may be solved by presenting the user with a certain, filtered, subset of the data. For example, a user might want to see data valid at a particular point in time, or may have permissions to view data only in a particular region. Rather than forcing application code to specify this filter criteria in tedious query conditions, Hibernate 3 will allow the (parametrized) filters to be applied at the session level.

(This feature has not yet been implemented, though some of the needed refactorings are currently sitting on my laptop.)

More Mapping Flexibility

For the overwhelming majority of cases, Hibernate2 provides all the O/R mapping options you will need. However, there are now a few new features that are powerful, if used judiciously.

  • single-class-to-multiple-table mappings using <join>
  • table-per-concrete-class-mappings using <union-subclass>
  • flexible discriminators using SQL formula mappings

We have long argued favor of extremely fine grained classes, and have viewed multi-table mappings as a huge misfeature (except for the very important special case of table-per-subclass inheritance mappings). Unfortunately, several commercial vendors insist upon trying to spin this misfeature as an actual competitive advantage of their products, causing us regular irritation. So we have provided the <join> mapping mainly to shut them up.

Well, I do conceed that there is one great use-case for <join>. We can now mix together table-per-hierarchy and table-per-subclass mappings in the same hierarchy. For example:

<class name="Document" table="DOCUMENTS">
    <id name="id">...</id>
    <discriminator column="TYPE" type="string"/>
    <subclass name="Book" discriminator-value="BOOK"/>
    <subclass name="Newspaper" discriminator-value="NEWSPAPER"/>
    <subclass name="XML"  discriminator-value="XML">
        <join table="XML_DOCUMENTS">

Hibernate's implicit polymorphism is a nice way to achieve most of the functionality of table-per-concrete-class without placing unnatural restrictions upon column types of different tables. It also nicely allows table-per-concrete-class to be mixed with other inheritance mapping strategies in the same hierarchy. Unfortunately, it has two limitations:

  • we cannot have a polymorphic collection of the superclass type
  • queries against the superclass resolve to muktiple SQL queries, which could be slower than using an SQL UNION.

The new <union-subclass> construct provides an explicit way to map classes to a table-per-concrete-class model and is implemented using SQL UNIONs.

If your table-per-class-hierarchy mapping does not feature a nice simple discriminator column, where values map one-to-one to the different classes, formula discriminators are for you. You can now discriminate between subclasses using multiple columns, multiple values of the same column, arbitrary SQL expressions or functions, even subqueries! An example:

<class name="Document" table="DOCUMENTS">
    <id name="id">...</id>
    <discriminator type="string"
        formula="case when TYPE='MONOGRAPH' then 'BOOK' when type='TABLOID'\
        or type='BROADSHEET' then 'NEWSPAPER' else TYPE end"/>
    <property name="type" column="TYPE"/>
    <subclass name="Book" discriminator-value="BOOK"/>
    <subclass name="Newspaper" discriminator-value="NEWSPAPER"/>

All of these features have been implemented in the Hibernate3 branch.

Representation Independence

Hibernate was conceived as a persistence solution for POJO domain models, and that remains the focus. Occasionally, we run into people who would like to represent their persistent entities in some more dynamic way, as a map, essentially. We used to point them to OFBiz Entity Engine. Well, Hibernate3 lets you represent your domain model as trees of HashMaps, or, with a little bit of user-written code, as just about anything. Here are three immediate applications of this new feature:

  • reimplement JBoss CMP 2.1 engine on top of Hibernate
  • reimplement OFBiz EntityEngine using Hibernate
  • natively persist SDOs

Indeed, the end goal is to allow the application to switch between whichever representation is appropriate to the task at hand; the same entity might be represented as a typesafe POJO, a Map, or an SDO, all with just a single Hibernate mapping.

This feature has been implemented in the Hibernate3 branch.

JDK 1.5 Support

JSR 175 annotations are a perfect fit for Hibernate metadata and we will embrace them aggressively. Emmanuel Bernard is working on this.

We will also need to support Java generics, which basically boils down to allowing typesafe collections (which is very trivial).

Stored Procedure Support

We are not enormous fans of using stored procedures for CRUD operations (of course, there are some other use cases where SPs make wonderful sense) but people working with legacy databases often need Hibernate to call a SP instead of generating its own SQL statement. In Hibernate 2.1, it is possible to achieve this with a custom persister. Hibernate3 will allow arbitrary SQL statements to be specified in the mapping document for create, update and delete. Max Andersen is currently implementing this feature.

Full Event Driven Design

In the early days of the project, I applied YAGNI aggressively. Hibernate was occasionally criticized for supposed architectural deficiencies, due to the absence of a clear upfront design. (In my view, this lack of upfront design was not actually an architectural problem - at least by my understanding of the word architecture - but that debate is for another day...) Anyway, it was my firm belief that an elegant and flexible design would eventually grow naturally, and that our scarce attention and development resources were better spent solving actual user-visible problems. I now feel quite vindicated in this decision; Hibernate has flourished, and the design that has emerged is both powerful and reasonably elegant. Take that, YAGNI-skeptics!

There is one final step of this process, that is now in the hands of Steve Ebersole (and already partially implemented). Hibernate3 will feature an event-oriented design, with event objects representing any interesting thing that occurs, and listener classes that implement standard Hibernate behaviours, or customized user-defined behaviors. This user extension capability has applications ranging for auditing to implementing strange cascade semantics. (Our previous attempt to solve these problems - the Interceptor interface - proved to be insufficient.) Importantly, the redesign simplifies the current monolithic SessionImpl class.

New AST-driven Query Parser

Well, actually /two/ final steps. (No-one expected the Spanish inquisition, did they?)

When I wrote the HQL parser, I knew zip about parsers. Hibernate ended up with a wierd-ass handwritten pure-look-behind parser which, to my great and enduring surprise, has actually served us very well and been the source of very few bugs. (YAGNI, once again.) But it is now well past time we had a proper ANTLR-grammar-based AST, and Joshua Davis is currently writing one. This probably won't mean much in the way of user-visible changes, but it should allow us to more easily support other query languages such as EJBQL.

Declarative Session Management

This is a JBoss-specific feature for now. New users often find session management to be tricky at first. It can take a bit of thought to get your head around the schizophrenic role played by the Hibernate session as both a kind of cache, and a kind of connection. A JDBC Connection is stateless; a Hibernate Session, on the other hand, is a stateful connection! We further confuse the issue by telling people that, actually, the session is really representing a kind of transaction!

Rather than force people to implement their own session handling, messing with threadlocals and exception handling, it would be nice to have some way to specify the session model (session-per-database-transaction, or session-per-application-transaction) declaratively, in the metadata for a session bean. The Spring Framework already provides some support for this kind of approach. Michael Gloegl is working on implementing this functionality using a JBoss Interceptor. It would be wonderful to have this facility in appservers other than JBoss. J2EE needs portable Interceptors for EJBs (they already exist for servlets).

Well, I have some more things on my list, but that will do for now!

SQL Tuning

Posted by    |       |    Tagged as

If you ever work with relational databases, you should go out and buy O'Reilly's /SQL Tuning/, by Dan Tow. The book is all about how to represent a SQL query in a graphical form and then, using some simple rules of thumb, determine an optimal execution plan for the query. Once you have found the optimal execution plan, you can add indexes, query hints, or use some other tricks to persuade your database to use this execution plan. Fantastic stuff. There is even sufficient introductory material for those of us (especially me) who know less than we should about the actual technical details of full table scans, index scans, nested loops joins, hash joins, etcetera to be able to start feeling confident reading and understanding execution plans. Unlike most database books out there, this book is not very platform-specific, though it does often refer specifically to Oracle, DB2 and SQL Server.

back to top