Spring Roo is a dynamic, domain-driven development framework from SpringSource, the makers of the insanely popular Spring framework, the de facto standard in enterprise Java. The Spring framework simplifies and expedites application development through a three-pronged approach: it enable services on plain-old-Java-objects (POJOs) declaratively and transparently through dependency injection and aspect-oriented programming, and—where functionality can’t be achieved effectively through those channels alone—it provides the simplest, cleanest abstractions and APIs under the sun to solve problems and to simplify existing, often verbose APIs.

If Spring’s so popular, and so productive, then surely Roo is redundant? After all, what could it possibly hope to add? “Spring’s the easiest way to work with Java today,” you think, “you just said it yourself!”

Spring is no doubt the most proficient way to work with Java, but the current thinking strongly supports the conclusion that the next barrier to enhancing productivity on the JMV is the Java language itself.

This too is not news.

SpringSource is also the custodian of the open-source Grails project, which has similar goals as Spring Roo. Grails is a highly productive web development framework built on the Groovy language. The framework’s built on top of Spring, but provides a workflow that’s far more like Ruby on Rails. Part of the productivity gains to be had in using this framework—part of its power—is that you can exploit the Groovy language’s dynamism and conciseness. Groovy’s a very dynamic language. It supports meta programming and the creation of exotic domain-specific languages. These features alone can pack quite a punch! They let the Grails developer specify more in far fewer lines of code than a Java developer could hope to achieve.

For some people, Grails is a compelling option, and the goal of Roo isn’t to take away from that. For some, however, Grails simply isn’t an option. Perhaps they can’t use Groovy in their environment, or they don’t want to make the large jump to Grails, feeling perfectly comfortable with their existing Spring and Java skills.

Figure 1-1. Pyramid of opinionation

Spring Roo is built using standard Java. You object, “But you just said … !” (I’m getting to that, hold on!) Spring Roo uses standard Java and Spring, but during development time, the Spring Roo shell watches you work, helping out as possible and required. Think of Spring Roo as being the ultimate pair-programming buddy, or the most advanced code completion you’ve ever seen.

As an example of this power, suppose you’re in the middle of editing a JPA entity in a Spring Roo project, and adding a field of interest—perhaps a dateOfBirth field to a Customer entity. As soon as you’ve finished typing out the field definition, Spring Roo automatically jumps in and adds a corresponding accessor and mutator pair for that field to a shadow class definition in the background. Similarly, it will implement a toString() definition (reflecting the fields added) if one does not already exist, and it will implement an equals() method following the same criteria. This assistance isn’t a one shot, either; it’s intelligent. If you update the field, the accessor and mutator are updated as well as the equals and toString methods. If you add an equals method to the JPA entity, the shadow definition is removed, delegating to your implementation instead. So, this shadow class definition is kept in sync, responding to your changes, but it does not get in your way. It defers to your will in all cases.

What is this shadow definition you ask? Well, it’s an AspectJ Inter Type Declaration (ITD) that Spring Roo maintains in the background. When your application compiles, the ITD is merged with the Java code, creating one class that has both the field you typed in, as well as the automatically generated accessor and mutator pair, a correct equals() implementation, and a correct toString() implementation. So, you write Java code, and Spring Roo augments the Java code with AspectJ Inter-Type Declarations (ITDs). You should never need to modify these ITD definitions. However, if you do decide to make modifications, do so at your own peril, as Spring Roo reserves the right to remove or modify them at will.

So, you get Java, but you don’t have to pay the cost of writing all that Java. For every line you write, Spring Roo will happily write any number of other boilerplate lines to relieve you of the burden. Because it is just code-generated Java and Spring, and nothing else, it’s as performant, well-written, and nicely architected as possible. Spring Roo is opinionated, but it always defers to you first.

Indeed, Spring Roo will never do anything unless you explicitly ask for it. It’s entirely opt-in—there’s no need to spend time undoing Spring Roo’s decisions. In the above example, we could bring Spring Roo into the project by explicitly directing it to do something for us from the shell, or by using compile-time retention-only annotations on our classes. If we did that, Spring Roo will monitor our workspace, using the annotations as a cue to intercede on our behalf, working in tandem with you in a background shadow definition.

Spring Roo’s very conducive to round tripping because of the aforementioned intelligence in the code generation. It’s possible to build your application entirely using Spring Roo. However, it may be that you eventually need to take your application out of Spring Roo’s sweet spot. Perhaps you’ve simply gotten 90% of your Spring application done, and want to take it the last 10% yourself. Here, too, Roo is different. You can completely remove Spring Roo from your application using push-in refactoring, yielding a generic Spring and Java-only web application that looks exactly as if you’d written it yourself and behaves exactly as it did when Spring Roo was managing its development.

This brings us around to the final piece of the puzzle: the development environment. After all, “refactoring” connotes IDEs, tooling, and Java development. And well it should! You already know that Spring Roo’s a shell that sits and monitors your code as you work, helping out wherever it can, passively, but what about these ITDs? The ITDs that it creates are not valid Java—they comply with the AspectJ language and can’t be compiled directly using javac. This is an easier problem to overcome than you might imagine. First, both Eclipse (in conjunction with the AspectJ Development Toolkit (ADJT) which is bundled with Eclipse and with SpringSource Tool Suite) as well as IntelliJ IDEA support the ITD format. As a result, when you work with Spring Roo projects, you still have access to code completion and to the refactoring support you’d expect. Those shadow definition accessors and mutators we created earlier will still show up in the code completion prompt in your favorite IDE. Additionally, every Spring Roo project ships with a correctly configured Maven build that automatically processes the code at compile time, so everything builds correctly.

Spring Roo is based on the sophisticated interplay between several moving parts. To do its work, Spring Roo needs to play a role during your development, and it must be there to help during compilation. This means you need a correctly configured development tool, and a correctly configured build process and test environment, beyond Spring Roo itself. This, as it turns out, is pretty easy to fix. In this section, we’ll set up the SpringSource Tool Suite (STS), a free development environment from SpringSource, based on Eclipse. The SpringSource Tool Suite’s got lots of extra features that makes working with Spring and the sister projects dead simple. Beyond being a particularly nice environment for Spring development, it’s also loaded to the gills with conveniences and useful-to-have packages. SpringSource Tool Suite always follows the main releases of Eclipse pretty closely, but integrates numerous plugins that can be a pain to set up independently, but that most people have to set up, anyway, like the Maven M2Eclipse plugin, or plugins for various source-code management options that aren’t included by default. It is effectively a subset of the Eclipse IDE for Java EE Developers, with a large superset of functionality integrated to reflect the realities of modern day enterprise Java development. We’ll use it throughout the book because it represents the path of least resistance and it’s a very capable choice, as well!

Because Spring Roo takes away so much of the boilerplate code, it is possible to develop Spring Roo applications using only a text editor and the Roo shell running in the background. Of course, we wouldn’t recommend it, but you could! Users of IntelliJ will be happy to know that the latest version of IntelliJ IDEA (the premium version) also supports Spring Roo development.

Let’s first obtain the SpringSource Tool Suite:

Figure 1-2. STS download page

Figure 1-3. Layout of STS folder

By default, the SpringSource Tool Suite comes integrated with Maven, Spring Roo, and the Developer Edition of SpringSource’s tcServer. SpringSource’s tcServer is a hardened, more robust distribution of the leading Apache Tomcat web server. The Developer Edition also includes integrated monitoring and analysis of your application in a package called Spring Insight, which lets you dissect the performance of your running Spring applications at fine granularity.

Let’s dig in by building something simple—you know, to kick the tires a bit.

Building a CRM is a rite of passage for every application developer today. We’ve all written one at some point or another. If pressed, we suspect most developers would admit to having done it—at least once (in college, perhaps? You don’t have to admit it aloud. You were young. It was a warm summer evening … the moon was full … and your awesome new startup (TM) needed a way to manage customer data. Totally cool. No worries. We won’t judge.)

Now that you’ve got SpringSource Tool Suite installed, fire it up, select a workspace, and then go to File > New > Spring Roo Project (Figure 1-4).

Figure 1-4. Opening a new Roo Project

The dialog’s a bit “busy,” but you don’t need to worry about most of it. This wizard simply dumps its inputs into a command line invocation, which you could do directly as well. Only fill out the first two fields—the rest you can leave as the defaults for the large majority of the time.

Figure 1-5. Creating a new Roo Project

Click “Next,” and then click “Finish.” After a flash of activity, this will dump you into the Spring Source Tool Suite with a fresh Spring project, based on Maven.

Figure 1-6. Spring Source Tool Suite with a fresh Roo Project

As you examine the workspace, you can see there’s a Roo Shell at the bottom of the screen. On the left is your freshly created project. First thing that you’ll notice is that it’s got a simple directory structure that confirms with the directory structure used in Maven builds. This project is a Maven project.

Maven projects defined their dependencies in a file called pom.xml that lives at the root of the project. Dependencies defined in pom.xml are automatically synchronized with the Eclipse project as classpath dependencies (you can inspect them in the project properties dialog or simply click on the “Maven Dependencies” Eclipse Classpath Container) thanks to the M2Eclipse plugin which comes bundled with SpringSource Tool Suite out of the box. As you enable features with Spring Roo, it will modify your pom.xml file and add dependencies as required. This is all automatic, and behind the scenes. The Maven file already represents a sophisticated build. A fresh Roo project already has JUnit, the correct and latest versions of Spring, AspectJ, logging, and the latest servlet APIs. Not bad! We’ll revisit this pom.xml as we build our application, so let’s get back to building our CRM.

Spring Roo applications are domain driven, everything you do stems from your domain objects, your entity objects. You describe your domain objects to Spring Roo and it will handle building persistence code (using JPA 2.0), web application code to manipulate those entities, and much more.

Don’t take my word for it, though. If you ever have a question, simply ask Spring Roo! Sometimes knowing that there is an answer is as good as knowing the answer. If you’re ever in doubt as to what Spring Roo can help you achieve, type “help,” and then hit Enter. Spring Roo will show all the commands that it can respond to. Similarly, if you ever have a doubt as to how to proceed, ask Roo by typing “hint” on the shell. Let’s do so:

Welcome to Spring Roo. For assistance press CTRL+SPACE or type "hint" then hit ENTER.

roo> hint
Roo requires the installation of a JPA provider and associated database.
Type 'persistence setup' and then hit CTRL+SPACE three times.
We suggest you type 'H' then CTRL+SPACE to complete "HIBERNATE".
After the --provider, press CTRL+SPACE twice for database choices.
For testing purposes, type (or CTRL+SPACE) HYPERSONIC_IN_MEMORY.
If you press CTRL+SPACE again, you'll see there are no more options.
As such, you're ready to press ENTER to execute the command.
Once JPA is installed, type 'hint' and ENTER for the next suggestion.
roo>

See? Everything’s about the domain objects—it wants you to set up the persistence infrastructure. So, oblige the shell—type in “persistence setup,” and then type CTRL + SPACE. If you’re using Roo from the shell inside of SpringSource Tool Suite, then the shell can offer you autocompletion using the same key commands as you’d use in the Java code editor. On a standard operating system shell, you’d use TAB, instead, just as you would to autocomplete commands in Bash, for example.

The code completion for the persistence setup command offers two options, both of which are required—one for --database, and one for --provider. Any code completion options that you get when you type CTRL + SPACE are required. Once you’ve selected a required code completion (like “persistence setup --database”), type CTRL + SPACE again to see if Roo has suggested values for that option. In this case, there are several different suggested values for --database. For expedience, I recommend you choose “H2_IN_MEMORY.” Hit CTRL + SPACE again, and the only remaining, required option (“--provider”), will be added to the command line. Type CTRL + SPACE again to get suggested values for the “--provider” option. Again, there are several options here, as well. One well-known option is “HIBERNATE” and unless you have a preference, go ahead and select it.

Often, commands may have optional arguments that you can use to fine tune the results. To see the optional arguments for any command, type “--,” and then type CTRL + TAB to see a full list of possibilities. Hit Enter.

The shell will flash and you’ll see telltale signs that Roo has added several new dependencies to your pom.xml in service of your request, including Hibernate, the Hibernate JPA 2.0 implementation, Commons Pool, Commons DBCP, the Hibernate JSR 303 (Java Bean Validation) implementation, and several Spring dependencies, including the Spring JDBC and transaction support. It’s also added a JPA persistence entity information file (src/main/resources/META-INF/persistence.xml), as well as Spring application context (in src/main/resources/META-INF/spring/applicationContext.xml). Finally, the Roo command has added a repository to the Maven pom.xml where new dependencies can be found.

Sure, the narrative of these last few paragraphs has taken more than a few minutes of your life, but the code—from nothing, to this point, has taken an almost trivial amount of effort. Let’s keep rolling and build our model.

The model is simple enough to be approachable, but complex enough that we can see how Spring Roo handles common scenarios. We won’t reprint all the Roo commands here, just enough so that you can see how to begin—the rest are easy to piece together.

Figure 1-7. The ER diagram for our domain model

Roo supports a few different workflows when building your domain objects using JPA. For greenfield domains, you can declaratively build your entities using the Spring Roo command line. We’ll build the model above using this approach. You should know that Spring Roo also supports reverse engineering existing domain models from a database (a feature called “database reverse engineering,” commonly abbreviated as “DBRE”) and generating the appropriate JPA entities. We’ll look at this approach shortly. Finally, there’s no reason you can’t mix ’n match these workflows: perhaps you will bootstrap your Spring Roo application using DBRE, but then use Spring Roo to maintain and evolve those entities.

To create an entity in Spring Roo, simply type “ent,” and then hit CTRL + TAB. It’ll auto-complete the command for you, filling in “entity” as well as the first and only required option, “--class.” Here you need to specify a class to use.

Recall that the first step of the process was to fill out the Spring Roo dialog, specifying a Project Name and a Top Level Package Name. The Top Level Package Name is the root package of your application. For instance, you might have a package, com.crmco.crm, that prefixes all parts of your application; the domain model lives in com.crmco.crm.model, the web-tier specific code lives in com.crmco.crm.web, etc. Roo seizes upon this convention and enshrines it in the idea of the Top Level Package Name. You can use the tilde character (“~”) as shorthand for the Top Level Package Name. It’ll be automatically replaced with the appropriate value whenever it’s referenced, in very much the same way as the tilde character is commonly substituted for the $HOME environment variable on Unixes.

So, for our Customer entity, simply specify “~.model.Customer” as the value for the --class option, and Roo will create a new JPA entity class com.crmco.crm.model.Customer.

Hit Enter and Roo will list the files that it has changed or created in response to your command:

roo> entity --class ~.model.Customer
Created SRC_MAIN_JAVA/com/crmco/crm/model
Created SRC_MAIN_JAVA/com/crmco/crm/model/Customer.java
Created SRC_MAIN_JAVA/com/crmco/crm/model/Customer_Roo_Configurable.aj
Created SRC_MAIN_JAVA/com/crmco/crm/model/Customer_Roo_Entity.aj
Created SRC_MAIN_JAVA/com/crmco/crm/model/Customer_Roo_ToString.aj
~.model.Customer roo>

A couple of things are striking about this output.

First, the prompt has changed. It was “roo>,” and now it’s “~.model.Customer roo>.” The prompt indicates the focus. The focus describes the subject of actions on the shell. In this case, Roo has given the entity you’ve just created focus. It knows that, more than likely, subsequent commands will attempt to modify this entity, and that it would be inconvenient if you had to specify the entity each time.

Second, Roo gave you four files, not just a single one! The first file (SRC_MAIN_JAVA/com/crmco/crm/model/Customer.java) you might correctly recognize as the Java class for your Customer JPA entity. It looks like this:

package com.crmco.crm.model;
import org.springframework.roo.addon.entity.RooEntity;
import org.springframework.roo.addon.javabean.RooJavaBean;
import org.springframework.roo.addon.tostring.RooToString;

@RooJavaBean
@RooToString
@RooEntity
public class Customer {
}

Kind of underwhelming, huh? Not a lot here. The only thing of interest here are the annotations at the top. These annotations are compile-time retention only—they disappear once the code’s compiled. They are used to signal to Roo that you want Roo to manage boilerplate code for you while you are developing. This boilerplate code exists in the other .aj files that were created.

You can click on the shell to open the other files, those ending in .aj. These files are the AspectJ ITDs we mentioned earlier. You won’t be able to see them in the Eclipse Package Explorer unless you disable a filter. To do so, click on the little down-arrow on the Package Explorer, choose Filters, and then find the checkbox for “Hide generated Spring Roo ITDs” (Figure 1-8). Uncheck it if it’s selected.

Figure 1-8. Disabling a filter

You are not supposed to modify these other files, they are there for Roo but lets see whats in them. Open the SRC_MAIN_JAVA/com/crmco/crm/model/Customer_Roo_Entity.aj file. This file exists because Spring Roo placed a @RooEntity annotation on your class. If you examine the code, you’ll see that Spring Roo’s already generated an ID for your Customer entity, it’s already generated a version column, and it’s already generated simple, ActiveRecord-style CRUD methods to create, remove, update, and delete a Customer object using a JPA EntityManager. If you examine the other file (Customer_Roo_ToSring.aj), you’ll see that Spring Roo knows about the instance variables in the class—including the ID and the version column—and already incorporates them into a useful toString() definition. The last file (Customer_Roo_Configurable.aj) simply adds the AspectJ @Configurable annotation to the entity so that Spring can transparently inject the JPA EntityManager into the entity class. Spring has no problems injecting collaborating beans into other Spring beans that it instantiates in the application context. Because Spring doesn’t control the construction and lifecycle of the JPA Customer entity—it needs to use AspectJ, which @Configurable specifically supports.

Let’s get back to it. We’ll define the column for the first name and the last name. To create the attribute, use the Roo field command, like this:

field string --fieldName lastName --notNull

Spring Roo will automatically add a field to the Customer.java class and it’ll generate corresponding accessors and mutators in the Customer_Roo_Entity.aj file. It also updates the Customer_Roo_ToString.aj to include the new field. Very smooth! You type a field, and Roo contorts to welcome that new field into the code base, updating everything as appropriate.

You could simply add the field to the entity class, if you’d like. In this case, it turns out to be roughly the same amount of typing so it really is a matter of choice. The Roo shell command has other options that you may not remember how to set up manually, which makes the Spring Roo command more convenient. For the lastName field, simply add it, manually:

@NotNull private String lastName;

Just as last time, Spring Roo hops into action, changing the other code as required to welcome the new field. See? Roo’s flexible. It gets out of your way and helps whenever it can.

So far, so good! Admittedly, this is a very simple entity so far. Let’s set up the CartOrder entity, which has a one-to-many relationship with the Customer entity. First, create the CartOrder entity, just as you did the Customer entity:

entity --class ~.model.CartOrder
field boolean --fieldName inCart
field date --fieldName dateOfOrder

To link the two, we need to use a special variant of the field command to establish the one-to-many relationship from the Customer, to the CartOrders. Let’s go back to our Customer entity and manipulate it.

Type:

focus --class ~.model.Customer

to change back to the Customer entity. Then, define the collection:

field set --fieldName orders --type ~.model.CartOrder --cardinality ONE_TO_MANY --mappedBy customer

This is half of our relationship, but we have to define the reverse side of the relationship, too. The “mappedBy” attribute tells Spring Roo that the field to use on the CartOrder entity is the customer attribute, presumably of type Customer. This is how you map a foreign key—the inverse side of this relationship—in JPA. Focus on the CartOrder, again, and then add the reference to the owning side—a Customer entity:

focus --class ~.model.CartOrder
field reference --fieldName customer --type ~.model.Customer --cardinality MANY_TO_ONE

It would be easy to continue in this way, fleshing out the balance of the model. Instead, however, let’s look at Spring Roo’s database reverse engineering feature (DBRE) to simply generate the model for us from an existing schema.

database reverse engineer --package ~.domain --schema PUBLIC

The command will fail because it can’t figure out which driver to use to connect to the database from which the domain model should be generated, but Spring Roo knew this might arise and has provided guidance. When you issue the command, it’ll respond that you need to install a driver for the database, and then show results:

database reverse engineer --package ~.domain --schema PUBLIC

Located add-on that may offer this JDBC driver
1 found, sorted by rank; T = trusted developer; R = Roo 1.1 compatible
ID T R DESCRIPTION ------------------------------------------------------------------------
01 Y Y 1.3.155.0020 H2 #jdbcdriver driverclass:org.h2.Driver. This bundle...
------------------------------------------------------------------------
[HINT] use 'addon info id --searchResultId ..' to see details about a search result
[HINT] use 'addon install id --searchResultId ..' to install a specific search result, or
[HINT] use 'addon install bundle --bundleSymbolicName TAB' to install a specific add-on version
JDBC driver not available for 'org.h2.Driver'

You heard the man … err … shell! Get hopping! It’s practically rolled the red carpet:

addon install id --searchResultId 01

Spring Roo will try to connect to the H2 database instance and look for tables to generate the entities from. It won’t find any, of course. You need to install the tables, which are available as part of this book’s source code download, as roo_crm.sql. In our local installation, we’re running the H2 database as a server, not in-memory. You can see what database properties Roo’s already using by using the database properties list command (which simply dumps the values in src/main/resources/META-INF/spring/database.properties).

Our local configuration looks like this:

database.password=
database.url=jdbc:h2:tcp://localhost/~/roo_crm
database.username=sa
database.driverClassName=org.h2.Driver

In this configuration, the H2 client will try to connect to an H2 database instance running as a server, locally. You can download the H2 database from www.h2database.com. Unzip it where you’d like and the run the h2.bat or h2.sh script inside the bin folder.

Change the H2 database connection URL like this:

database properties set --key database.url --value jdbc:h2:tcp://localhost/~/roo_crm

It’ll churn for a second, and then return you to the shell, at which point you can simply re-issue the database reverse engineering command. Note that we’ve specified ~.domain, not ~.model, to avoid having the DBRE command overwrite, and conflict with, the existing Customer class definition. You may simply delete the existing ~.model package.

The entities that are brought in from the DBRE process are empty—all the fields that were generated are in an ITD—of the form $ENTITY_Roo_DbManaged.aj.

If you’re confident in the reverse engineering, then open up the $ENTITY_Roo_DbManaged.aj file and move all the fields into the entity directly. This will cause the $ENTITY_Roo_DbManaged.aj file to be deleted, and works more naturally when adding finder methods, as we describe in the next section.

Thus far, we’ve simply built our entities and relied on the generated query methods that are added to them by default. We can, for example, interact with our database using the methods added to each of our entities by default:

// create a new Customer entity
Customer customer = new Customer() ;
customer.setFirstName("steve");
customer.setLastName("mayzak");
customer.persist();

// presumably we've gotten ahold of the ID
Customer steve = Customer.findCustomer( 1L ) ;
steve.setFirstName( "Steve" );
steve.setLastName( "Mayzak" );
steve.merge() ;

// count the records in the database
long customers = Customer.countCustomers();

This is a start, but often we need to retrieve data using more sophisticated queries. To do this in Spring Roo, we need to create “finder” methods. This, as it turns out, is very easy. Spring Roo has a whole arsenal of finders that it knows how to generate without your help at all. First consult this list to see if any of the ones it can add are what you’re looking for:

~.domain.Customer roo> focus --class ~.domain.Customer
~.domain.Customer roo> finder list
findCustomersByCartOrders(Set cartOrders)
findCustomersByCompanyId(Company companyId)
findCustomersByFirstNameEquals(String firstName)findCustomersByFirstNameIsNotNull()
findCustomersByFirstNameIsNull()
findCustomersByFirstNameLike(String firstName)
findCustomersByFirstNameNotEquals(String firstName)
findCustomersByLastNameEquals(String lastName)
findCustomersByLastNameIsNotNull()
findCustomersByLastNameIsNull()
findCustomersByLastNameLike(String lastName)
findCustomersByLastNameNotEquals(String lastName)
~.domain.Customer roo>

Use the finder add command, specifying the finder name as specified in the finder list output:

~.domain.Customer roo> finder add --finderName findCustomersByFirstNameLike

Open up the newly created Customer_Roo_Finder.aj file to inspect the finder method that was created for you.

If you don’t see a finder that you want already, you can of course simply add a finder as you’d like in the entity’s Java class, in the same style as the finders generated by Spring Roo.

Spring Roo builds active record style entity objects. Active record objects—from Martin Fowler’s active record pattern—are essentially entity objects whose interface includes methods to store, retreive (through queries, or by ID), update, and delete the entity. While the active record pattern is famously used quite succesfully in Ruby on Rails and Grails, it’s little known in typical enterprise Java shops where the preferred approach is to use services, or, as often, services and repositories. It is still easy to build a service tier facade with Spring Roo, delegating to the active record-style entities to perform basic repository object duties.