O'Reilly logo

Continuous Enterprise Development in Java by Aslak Knutsen, Andrew Lee Rubinger

Stay ahead with the world's most comprehensive technology and business learning platform.

With Safari, you learn the way you learn best. Get unlimited access to videos, live online training, learning paths, books, tutorials, and more.

Start Free Trial

No credit card required

Chapter 4. Scratch to Production

“The way to get started is to quit talking and begin doing.” - Walt Disney

Enterprise Java has long suffered the (possibly correct) critique that it’s difficult to bootstrap a new project. Couple the lack of definitive jumpstart documentation with vendor-specific techniques for application deployment, throw a mess of 3rd-party dependencies into the mix, and we’ve got a prime recipe yielding barriers to entry for programmers new to web development in Java.

Of course, this all runs contrary to the mission of Java EE: to make our experience with enterprise features easier. So while the programming model has certainly evolved past the days of confusingly verbose and explicitly-required metadata, the warts which lead to frustrating stack traces and unexpected deployment behaviors unfortunately persist.

Some of this is by design. The Specifications which comprise the Java EE Platform intentionally leave room for vendors to implement features like server startup and deployment at their discretion [1].

In the interest of providing a uniformly-workable solution to the reader, this text will routinely opt for vendor-specific approaches in favor of generic guidelines. By the end of this chapter, you should be comfortable creating a new Java EE Web Application and pushing it live to production using a few tools and services offered by the JBoss Community.

The Development Environment

While all projects used here are ultimately standalone and require no plugins or special environments aside from a Java runtime, we’re going to make our lives easier by taking advantage of the integration facilities provided by JBoss Developer Studio (JBDS).

JBoss Developer Studio(TODO: Get an updated approved image)

The JBDS plugins atop the Eclipse Integrated Development Environment (IDE) will unify our development experience and allow us to stay inside one window. Installation is via an executable JAR available from the JBDS Site TODO Examine link and ensure it’s up-to-date before final publication.

To kick off the installation process, either double-click the icon (if your environment has the “.jar” extension correctly associated as a Java executable) or launch the installer from the command line via the Java runtime:

$> java -jar jbdevstudio-product-universal-6.0.0.GA-v20121206-1855-B186.jar

JBoss Developer Studio Installation(TODO: Ensure version is correct)

Following the graphical wizard will install the JDBS (and all requisite plugins we’ll be using) IDE onto your local machine.

A New Project

The previous chapter introduced us to JBoss Forge, a tool that aims to make project creation and enhancement more declarative and less manual. As we’re starting fresh now, it makes sense to use Forge to create our project layout. This will ultimately give us a functional skeleton from database to view layer which we can use either as a learning tool or a quick shortcut to writing some real code.

Forge’s user interface is a shell, so it can be installed manually and used from the terminal like any other command-line application. However, JBDS removes the need for us to do this setup. Selecting Window > Show View > Other… will give us immediate access to the Forge Console:

Forge Console View Selection

With our new Forge Console view, we’re now free to start up the Forge runtime, which came embedded with the JBDS installation. Pressing the green “Play” button will give us access to the Forge shell.

Start Forge

   |  ___|__  _ __ __ _  ___
   | |_ / _ \| `__/ _` |/ _ \  \\
   |  _| (_) | | | (_| |  __/  //
   |_|  \___/|_|  \__, |\___|

JBoss Forge, version [ 1.0.6.Final ] - JBoss, by Red Hat, Inc. [ http://jboss.org/forge ]
[no project] pwd $

JBDS integration with Forge is especially useful in this console as the IDE will automatically refresh any changes we make in Forge with our project view and open text editors.

As a decent shell, Forge support tab-complete of commands and known parameters; if you get stuck, feel free to use the TAB key to see what’s available.

To ease up on our configuration options, let’s first start off by instructing Forge to accept defaults:

$> set ACCEPT_DEFAULTS true;

And now let’s create the filesystem layout and pom.xml for our new Maven-based Java EE project. We’ll be creating a simple application which will allow users to leave comments, so we’ll name the application, “feedback”:

$> new-project --named feedback --topLevelPackage org.cedj.ch03.feedback --projectFolder feedback;

Once we hit enter, we’ll see that Forge has dutifully created our new project’s layout:

***SUCCESS*** Created project [feedback] in new working directory [./feedback]
Wrote ./feedback
Wrote ./feedback/pom.xml
Wrote ./feedback/src/main/java
Wrote ./feedback/src/test/java
Wrote ./feedback/src/main/resources
Wrote ./feedback/src/test/resources
Wrote ./feedback/src/main/java/org/cedj/feedback
Wrote ./presentations/feedback/src/main/resources/META-INF/forge.xml

Additionally, our project has appeared in the Project View:

Project Created

Users of Maven Archetypes may be familiar with this type of technique to create a new project, but as Forge is an incremental tool, it’s capable of reading a project’s state and adding behaviors after creation.

Let’s add support for Java Persistence (JPA) to our project, a task that typically would involve some searching for the correct dependencies for the spec APIs (as well as those for any vendor-specific extensions). Forge is helpful here as well, via its persistence plugin:

$> persistence setup --provider HIBERNATE --container JBOSS_AS7;

In this case we’ve chosen Hibernate as our persistence provider, and have targeted JBoss AS7 as our container. Forge will equip our POM with the proper dependencies and supply us with a default persistence.xml preconfigured to work with the AS7 runtime. Remember, for a list of supported options, look to TAB completion.

***SUCCESS*** Installed [forge.spec.jpa] successfully.
***INFO*** Setting transaction-type="JTA"
***INFO*** Using example data source [java:jboss/datasources/ExampleDS]
***SUCCESS*** Persistence (JPA) is installed.
Wrote ./feedback/src/main/resources/META-INF/persistence.xml
Wrote ./feedback/pom.xml

A peek into the generated persistence.xml will show us a decent default configuration:

<?xml version="1.0" encoding="UTF-8" standalone="no"?>
<persistence xmlns="http://java.sun.com/xml/ns/persistence" xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" version="2.0" xsi:schemaLocation="http://java.sun.com/xml/ns/persistence http://java.sun.com/xml/ns/persistence/persistence_2_0.xsd">
  <persistence-unit name="forge-default" transaction-type="JTA">
    <description>Forge Persistence Unit</description>
      <property name="hibernate.hbm2ddl.auto" value="create-drop"/>
      <property name="hibernate.show_sql" value="true"/>
      <property name="hibernate.format_sql" value="true"/>
      <property name="hibernate.transaction.flush_before_completion" value="true"/>

Let’s make one tweak; the property hibernate.hbm2ddl.auto is set to automatically drop the database tables such that they won’t be able to be reused across deployments. While this might be handy in development to ensure you’re always coding from a clean slate, we’d actually like to use some real persistence later on, so let’s change that property to a value of update.

Java EE6 introduced the Bean Validation Specification which allows for validation constraints at the database, application, and view layers all with a single declaration. Let’s enable BV for our project, similar to how we put in place support for persistence:

$> validation setup --provider HIBERNATE_VALIDATOR

Once again we’re given the appropriate dependencies in our POM, as well as a valid validation.xml configuration file such that we don’t have to apply any boilerplate XML on our own.

***SUCCESS*** Installed [forge.spec.validation] successfully.
Wrote ./feedback/src/main/resources/META-INF/validation.xml
Wrote ./feedback/pom.xml

The generated validation.xml should be fine for our uses without any modification.

<?xml version="1.0" encoding="UTF-8" standalone="no"?>
<validation-config xmlns="http://jboss.org/xml/ns/javax/validation/configuration" xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance">

Now we’re all set to add some entities to our project. For the uninitiated, this will be our interface to accessing persistent (ie. database-backed) data as an object. For now we’ll just create one simple bean to represent a database table, and we’ll call it “FeedbackEntry”.

$> entity --named FeedbackEntry;

Forge will create a new Java class for us, adding the proper @Entity annotation, an ID field to represent our primary key, a version field for optimistic locking, and stubbed out methods for value-based equals(Object) and hashCode().

package org.cedj.feedback.model;

import javax.persistence.Entity;
import java.io.Serializable;
import javax.persistence.Id;
import javax.persistence.GeneratedValue;
import javax.persistence.GenerationType;
import javax.persistence.Column;
import javax.persistence.Version;
import java.lang.Override;

public class FeedbackEntry implements Serializable

   private @GeneratedValue(strategy = GenerationType.AUTO)
   @Column(name = "id", updatable = false, nullable = false)
   Long id = null;
   private @Column(name = "version")
   int version = 0;

   public Long getId()
      return this.id;

   public void setId(final Long id)
      this.id = id;

   public int getVersion()
      return this.version;

   public void setVersion(final int version)
      this.version = version;

   public String toString()
      String result = "";
      if (id != null)
         result += id;
      return result;

   public boolean equals(Object that)
      if (this == that)
         return true;
      if (that == null)
         return false;
      if (getClass() != that.getClass())
         return false;
      if (id != null)
         return id.equals(((FeedbackEntry) that).id);
      return super.equals(that);

   public int hashCode()
      if (id != null)
         return id.hashCode();
      return super.hashCode();

Our FeedbackEntry entity should be capable of recording feedback for some user with a Twitter ID, so let’s add fields to represent that data (as well as some validation constraints dictating that these may not be null).

field string --named twitterHandle;
constraint NotNull --onProperty twitterHandle;
field string --named feedback;
constraint NotNull --onProperty feedback;

It’s worth noting now that our Forge prompt reads that the current location is inside our Entity, as that’s where we’re currently working. Forge’s ls command is handy for seeing the current state of our Entity as we build.

[feedback] FeedbackEntry.java $ ls


public::equals(Object that)::boolean
public::setFeedback(final String feedback)::void
public::setId(final Long id)::void
public::setTwitterHandle(final String twitterHandle)::void
public::setVersion(final int version)::void

With our sole Entity in place, it’s time to let Forge generate a UI layer for us as a starting point for the view in our web application. The "scaffolding" command makes short work of this.

$> scaffold setup
***SUCCESS*** Installed [forge.maven.WebResourceFacet] successfully.
***SUCCESS*** Installed [forge.spec.ejb] successfully.
***SUCCESS*** Installed [forge.spec.cdi] successfully.
***SUCCESS*** Installed [forge.spec.servlet] successfully.
***SUCCESS*** Installed [forge.spec.jsf.api] successfully.
***SUCCESS*** Installed [faces] successfully.
Wrote ./feedback/src/main/webapp
Wrote ./feedback/pom.xml
Wrote ./feedback/src/main/webapp/WEB-INF/beans.xml
Wrote ./feedback/src/main/webapp/WEB-INF/faces-config.xml
Wrote ./feedback/src/main/webapp/favicon.ico
Wrote ./feedback/src/main/webapp/resources/scaffold/paginator.xhtml
Wrote ./feedback/src/main/webapp/resources/scaffold/pageTemplate.xhtml
Wrote ./feedback/src/main/webapp/index.html
Wrote ./feedback/src/main/webapp/index.xhtml
Wrote ./feedback/src/main/webapp/error.xhtml
Wrote ./feedback/src/main/webapp/resources/add.png
Wrote ./feedback/src/main/webapp/resources/bootstrap.css
Wrote ./feedback/src/main/webapp/resources/false.png
Wrote ./feedback/src/main/webapp/resources/favicon.ico
Wrote ./feedback/src/main/webapp/resources/forge-logo.png
Wrote ./feedback/src/main/webapp/resources/forge-style.css
Wrote ./feedback/src/main/webapp/resources/remove.png
Wrote ./feedback/src/main/webapp/resources/search.png
Wrote ./feedback/src/main/webapp/resources/true.png
Wrote ./feedback/src/main/webapp/WEB-INF/web.xml

As shown by the somewhat lengthy output, we’re now equipped with a src/main/webapp folder laid out with a nice starting point from which we can build our own UI. With just one more command, we can generate a CRUD (Create, Read, Update, Delete) interface to our entities:

$> scaffold from-entity org.cedj.feedback.model.*;
***INFO*** Using currently installed scaffold [faces]
***SUCCESS*** Generated UI for [org.cedj.feedback.model.FeedbackEntry]
Wrote ./feedback/src/main/java/org/cedj/feedback/view/FeedbackEntryBean.java
Wrote ./feedback/src/main/webapp/feedbackEntry/create.xhtml
Wrote ./feedback/src/main/webapp/feedbackEntry/view.xhtml
Wrote ./feedback/src/main/webapp/feedbackEntry/search.xhtml
Wrote ./feedback/src/main/webapp/resources/scaffold/pageTemplate.xhtml
Wrote ./feedback/src/main/java/org/cedj/feedback/view/ViewUtils.java
Wrote ./feedback/src/main/webapp/WEB-INF/classes/META-INF/forge.taglib.xml
Wrote ./feedback/src/main/java/org/cedj/feedback/model/FeedbackEntry.java

And that’s enough for now; we’ve created the skeleton for a fully-functional application. Of course, the thematic element of this book is testable development, so it’s best we throw in the facility to run some integration tests on our little application.

Writing Our First Integration Test with Arquillian

We’ve mentioned before that Forge is based on a plugin architecture; all commands we’ve used thus far are actually plugins called by the Forge runtime when we request them in the console. Up to this point, we’ve used support that comes standard with the Forge distribution. Now we’d like to add some tests, and we’ll use the Arquillian Test Platform as both the programming model and the JUnit test runner. First order of business is to install the Arquillian plugin into our Forge runtime, and this is done by way of the forge install-plugin command.

$> forge install-plugin arquillian
Connecting to remote repository [https://raw.github.com/forge/plugin-repository/master/repository.yaml]... connected!
***INFO*** Preparing to install plugin: arquillian
***INFO*** Checking out plugin source files to [/tmp/forgetemp1365281623326595751/repo] via 'git'
***INFO*** Switching to branch/tag [refs/heads/1.0.2.Final]
***INFO*** Invoking build with underlying build system.
***INFO*** Installing plugin artifact.
***SUCCESS*** Installed from [https://github.com/forge/plugin-arquillian.git] successfully.

This instructs Forge to connect to its plugin repository, grab the latest version of the requested plugin, build it from source, and install the binaries into the current runtime. As Forge is built on a modular ClassLoading architecture, we’re able to load in plugins without the need to restart the process or concern ourselves with conflicting dependencies.

With the Arquillian plugin installed, we now have access to the arquillian command. Let’s instruct Forge to equip our POM with the dependencies needed to run Arquillian tests on the JBoss AS7 container.

$> arquillian setup --container JBOSS_AS_REMOTE_7.X;

You’ll be prompted for the versions of Arquillian, JUnit, and JBoss AS7 that you’d like to use, and the available options will expand over time as new versions are released. These instructions have been tested with:


With the POM config changes out of the way, let’s ask Forge to now create for us a jumping-off point from which we’ll write our test.

$> arquillian create-test --class org.cedj.feedback.model.FeedbackEntry.java
Picked up type <JavaResource>: org.cedj.feedback.model.FeedbackEntryTest
Wrote ./feedback/src/test/java/org/cedj/feedback/model/FeedbackEntryTest.java

The newly-created FeedbackEntryTest is technically an Arquillian test, but it really doesn’t do too much for us. After all, we can automate quite a bit, but in the end it’s up to us to write our own business and test logic. So let’s replace the contents of this class with:

package org.cedj.feedback.model;

import java.io.File;
import javax.persistence.EntityManager;
import javax.persistence.PersistenceContext;
import org.jboss.arquillian.container.test.api.Deployment;
import org.jboss.arquillian.junit.Arquillian;
import org.jboss.shrinkwrap.api.ShrinkWrap;
import org.jboss.shrinkwrap.api.spec.WebArchive;
import org.junit.Assert;
import org.junit.Test;
import org.junit.runner.RunWith;

public class FeedbackEntryTest {
    private EntityManager em;

    public static WebArchive createDeployment() {
        return ShrinkWrap.createFromZipFile(WebArchive.class, new File(

    public void canFindFeedbackByUser() {
        final FeedbackEntry feedback = em.createQuery(
                "from " + FeedbackEntry.class.getSimpleName()
                        + " where twitterHandle='@ALRubinger'",

    public void testIsDeployed() {

Before going forward, let’s break down the anatomy of this test.

First, we’ll note that there are no references in the import statements to any particular application server or target container. This is because Arquillian is designed to decouple the programming model of the test from the target runtime; any container which can handle the capabilities demanded by the test will work. This keeps the portability goals of Java EE intact, moving the mechanics of startup and deployment to configuration elements. In this case, the Arquillian runner will note that the JBossAS7 container adaptor is available on the classpath as it’s been defined in the POM when we ran the setup command for the Arquillian Forge plugin.

The next point of interest is the class-level annotation:


@RunWith is a standard JUnit construct which directs control to a specified test runner. This is Arquillian’s entry point; from here Arquillian can receive lifecycle events from JUnit and perform its own handling. The benefit to this design decision is that Arquillian requires no special plugins or configuration on the part of the user. Anything which is capable of launching a JUnit test - be it a Maven build, an Ant task, a manual command, or an IDE - can take advantage of Arquillian without any additional handling. For instance, JBDS and Eclipse can launch a full-scale integration test with Arquillian by right-clicking on the class and selecting "Run As > JUnit Test“.

Next up is the class declaration:

public class FeedbackEntryTest {...}

The important bit here is what’s not required. Because of the Arquillian JUnit Test Runner, you’re free to use whatever class hierarchy you’d like, and there’s no need to extend a base support class. This keeps Arquillian tests in line with the POJO programming model originally introduced in Java EE5.

Another feature of Arquillian is its ability to provide services like injection to the test. Here we’re going to interact with persistent storage via the JPA EntityManager:

    private EntityManager em;

The EntityManager is typically used by server-side business components like EJBs or CDI beans, but because this test is going to run inside the container as part of a deployment, we’ll be able to interact with it directly.

TODO: Insert image of the Arquillian lifecycle with regards to in-container testing

Because Arquillian aims to follow the standards set forth by Java EE, instead of requiring the user to do a lookup or manual creation of the EntityManager, we’ll be able to receive an instance by requesting injection via use of the @PersistenceContext annotation.

The final important fixture of the Arquillian test anatomy is the @Deployment method:

    public static WebArchive createDeployment() {
        return ShrinkWrap.createFromZipFile(WebArchive.class, new File(

Because Java EE application servers work off deployments like Web Archives (WARs), Java Archives (JARs), or Enterprise Archives (EARs), we need to instruct Arquillian with the artifact to be deployed. This method must be static and return any ShrinkWrap Archive type; for this first exercise we’ll simply grab the output of the current project’s build feedback.war, but as we’ll soon see in later examples, we don’t need to rely on flat files at all! This will free us to skip the build entirely inbetween code changes and test runs, instead letting us rely ShrinkWrap’s packaging of .class files created from the IDE’s incremental complication features.

The rest of the file is all test logic! Remember, the focus of the Arquillian programming model is to allow you to write less boilerplate and setup, and focus on the bits of code that only you as the developer can write. It’s not your job to deal with bootstrapping an application server or calling upon vendor-specific deployment hooks; Arquillian will handle all of that for us behind the scenes.

Running the Application Locally

Time to see our generated application in action. First we should run the build to package our flat-file deployable feedback.war for manual deployment into JBoss AS7. We can trigger Maven from the Forge console:

$> build --notest --profile JBOSS_AS_REMOTE_7.X;

After a series of informative build output messages from Maven, you should see BUILD SUCCESS, indicating that the WAR has been properly built from sources.

The missing bit is that we need a server into which we can deploy our webapp! JBossAS7 has a simple installation process (simply download and unzip onto the filesystem), but again Forge can help automate this for us so we don’t need to locate the JBossAS binaries. For this we’ll turn to the Forge JBossAS7 Plugin, which is installed similarly to the Arquillian plugin we put in place in the last section.

$> forge install-plugin jboss-as-7

Once installation is complete, we may use the newly-acquired as7 command to set up our server.

$> as7 setup

You’ll be prompted for your $JAVA_HOME location and JBossAS7 version; be sure to align the versions with the Arquillian Container Adaptor Version we’d chosen before. Again, in this example we recommend 7.1.1.Final. Forge will additionally ask for the location to a JBossAS7 installation on the filesystem, but simply hitting ENTER will download the server for us into the target directory of our project.

Now it’s time to fire up the server. First cd into the root of your project in the Forge shell, then execute:

$> as7 start --jboss-home target/jboss-as-dist/jboss-as-7.1.1.Final/

If you’ve opted for a different version of JBossAS7, you may have to make substitutions to point JBOSS_HOME correctly. Assuming all goes to plan, you should see the JBossAS7 startup sequence in the Forge shell, followed by:

***INFO*** JBoss AS 7.1.3.Final has successfully started.

With the server up, let’s deploy our application:

$> as7 deploy

Again, after a series of JBossAS7 deployment messages, you should see:

The deployment operation (FORCE_DEPLOY) was successful.

We’re up and running! Point your browser of choice to the root of the application at http://localhost:8080/feedback, and you should see the home screen of the UI that Forge has generated for us.

Feedback Application Home

Selecting the “Feedback Entry” button will grant us access to the CRUD editor for this entity. From here we can create a new row in the database table.

New Feedback Entry

While CRUD applications are little more than a UI frontend to an Entity, the benefit here is in having a fully-functioning application to use as a base from which to start. For newcomers to Java EE, this is especially useful as a learning tool.

With our new entry now persisted into the database, let’s undeploy the application in preparation to perform our first integration test run with Arquillian.

$> as7 undeploy
The deployment operation (UNDEPLOY_IGNORE_MISSING) was successful.

Running the Arquillian Integration Test

At this point, we still have a running JBoss AS7 server and have undeployed the “feedback” application. Because we’d chosen the JBOSS_AS_REMOTE_7.X option as part of the Forge Arquillian Plugin setup command above, our POM is equipped with a profile which enables a dependency on the JBoss AS7 Arquillian Container:


Let’s inform JBDS that we should consider the metadata considered in this profile; this will impact our compilation and JUnit runtime +classpath+s.

Select Maven Profile

Now the Arquillian test launcher will know to pick up the proper adaptor to a remote JVM instance of JBoss AS7 when running tests; it will connect to the currently-running instance, deploy the defined @Deployment, execute the tests, and undeploy to clean up. If we’d like to allow Arquillian to automatically control the server start/stop lifecycle alongside each test suite, we could alternatively use the JBOSS_AS_REMOTE_7.X setup option which defines org.jboss.as:jboss-as-arquillian-container-managed as a dependency in a POM profile.

With JBDS now configured with the proper classpath for test execution, all that’s left to do is launch the test. A simple right-click on the test class in the Project Explorer yields the option Run As > JUnit Test. The IDE’s JUnit launcher will create a new process, fire up JUnit, and yield control to Arquillian. We’ll receive results just as we’d expect from any other JUnit test.

Passing the Tests

With assurance that our application has some minimal level of tested functionality, let’s take a risk and move this off the isolation of our local machine and into the public realm, accessible from the world.

[1] While there is some limited facility to, for instance, create an EJB container in a running JVM and bring EJB deployments on the classpath into service, a full-scale deployment is still typically achieved in a vendor-specific manner

With Safari, you learn the way you learn best. Get unlimited access to videos, live online training, learning paths, books, interactive tutorials, and more.

Start Free Trial

No credit card required