rmi

The Rmi plugin adds a remoting client that uses the Java RMI protocol. It is compatible with Grails' Remoting plugin 1.3.

Usage

The plugin will inject the following dynamic methods:

• <R> R withRmi(Map<String, Object> params, Closure<R> stmts) - executes stmts using a RmiClient
• <R> R withRmi(Map<String, Object> params, CallableWithArgs<R> stmts) - executes stmts using a RmiClient

Where params may contain

All dynamic methods will create a new client when invoked unless you define an id: attribute. When this attribute is supplied the client will be stored in a cache managed by the RmiProvider that handled the call. The plugin will attempt to locate the default RmiRegistry when the lazy: property is set to false.

These methods are also accessible to any component through the singleton griffon.plugins.rmi.RmiEnhancer. You can inject these methods to non-artifacts via metaclasses. Simply grab hold of a particular metaclass and call RmiEnhancer.enhance(metaClassInstance).

Configuration

RmiAware AST Transformation

The preferred way to mark a class for method injection is by annotating it with @griffon.plugins.rmi.RmiAware. This transformation injects the griffon.plugins.rmi.RmiContributionHandler interface and default behavior that fulfills the contract.

Dynamic Method Injection

Dynamic methods will be added to controllers by default. You can change this setting by adding a configuration flag in griffon-app/conf/Config.groovy

griffon.rmi.injectInto = ['controller', 'service']

Dynamic method injection will be skipped for classes implementing griffon.plugins.rmi.RmiContributionHandler.

Example

This example relies on Grails as the service provider. Follow these steps to configure the service on the Grails side:

1. Download a copy of Grails and install it.

2. Create a new Grails application. We'll pick 'exporter' as the application name.

   grails create-app exporter
   

3. Change into the application's directory. Install the remoting plugin.

   grails install-plugin remoting
   

4. Create the following interface in src/groovy/exporter/Calculator.groovy. This interface will be used on the Griffon side too.

   package exporter
   import java.rmi.Remote
   import java.rmi.RemoteException
   interface Calculator extends Remote {
       double add(double a, double b) throws RemoteException
   }
   

5. Create an implementation of the Calculator interface as a service

   grails create-service calculator
   

6. Paste the following code in grails-app/services/exporter/CalculatorService.groovy

   package exporter
   class CalculatorService implements Calculator {
       boolean transactional = false
       static expose = ['rmi']
   
       double add(double a, double b) {
           println "add($a, $b)" // good old println() for quick debugging
           return a + b
       }
   }
   

7. Run the application

   grails run-app
   

Now we're ready to build the Griffon application

1. Create a new Griffon application. We'll pick calculator as the application name

   griffon create-app calculator
   

2. Install the rmi plugin

   griffon install-plugin rmi
   

3. Fix the view script to look like this

   package calculator
   application(title: 'Rmi Plugin Example',
     pack: true,
     locationByPlatform: true,
     iconImage: imageIcon('/griffon-icon-48x48.png').image,
     iconImages: [imageIcon('/griffon-icon-48x48.png').image,
                  imageIcon('/griffon-icon-32x32.png').image,
                  imageIcon('/griffon-icon-16x16.png').image]) {
       gridLayout(cols: 2, rows: 4)
       label('Num1:')
       textField(columns: 20, text: bind(target: model, targetProperty: 'num1'))
       label('Num2:')
       textField(columns: 20, text: bind(target: model, targetProperty: 'num2'))
       label('Result:')
       label(text: bind{model.result})
       button(calculateAction, enabled: bind{model.enabled})
   }
   

4. Let's add required properties to the model

   package calculator
   @Bindable
   class CalculatorModel {
       String num1
       String num2
       String result
       boolean enabled = true
   }
   

5. Now for the controller code. Notice that there is minimal error handling in place. If the user types something that is not a number the client will surely break, but the code is sufficient for now.

   package calculator
   @griffon.plugins.rmi.RmiAware
   class CalculatorController {
       def model
   
       def calculate = { evt = null ->
           double a = model.num1.toDouble()
           double b = model.num2.toDouble()
           execInsideUISync { model.enabled = false }
           try {
               def result = withRmi(host: 'localhost', port: 1199) {
                   service('exporter.Calculator') {
                       add(a, b)
                   }
               }
               execInsideUIAsync { model.result = result.toString() }
           } finally {
               execInsideUIAsync { model.enabled = true }
           }
       }
   }
   

6. Locate the compiled classes from Grails; jar the calculator interface and place it in the lib directory of the Griffon application. Assume $grailsProject points to the directory of the exporter application and $griffonProject points to the calculator application

   cd $grailsProject/target/classes
   jar cvf $griffonProject/lib/exporter-api.jar exporter/Calculator.class
   

7. Run the application

   griffon run-app
   

The plugin exposes a Java friendly API to make the exact same calls from Java, or any other JVM language for that matter. Here's for example the previous code rewritten in Java. Note the usage of @RmiWare on a Java class

    package calculator;
    import griffon.util.CallableWithArgs;
    import griffon.util.CollectionUtils;
    import griffon.plugins.rmi.RmiClient;
    import java.awt.event.ActionEvent;
    import java.util.Map;
    import org.codehaus.griffon.runtime.core.AbstractGriffonController;
    @griffon.plugins.rmi.RmiAware
    public class CalculatorController extends AbstractGriffonController {
        private CalculatorModel model;

        public void setModel(CalculatorModel model) {
            this.model = model;
        }

        public void calculate(ActionEvent event) {
            final double a = Double.parseDouble(model.getNum1());
            final double b = Double.parseDouble(model.getNum2());
            enableModel(false);
            try {
                Map<String, Object> params = CollectionUtils.<String, Object> map()
                      .e("host", "localhost")
                      .e("port", 1199);
                final Double result = withRmi(params,
                    new CallableWithArgs<Double>() {
                        public Double call(Object[] args) {
                            RmiClient client = (RmiClient) args[0];
                            return (Double) client.service("Calculator",
                                new CallableWithArgs<Double>() {
                                    public Double call(Object[] args2) {
                                        return ((Calculator) args2[0]).add(a, b);
                                    }
                                });
                        }
                    });
                execInsideUIAsync(new Runnable() {
                    public void run() {
                        model.setResult(String.valueOf(result));
                    }
                });
            } finally {
                enableModel(true);
            }
        }

        private void enableModel(final boolean enabled) {
            execInsideUIAsync(new Runnable() {
                public void run() {
                    model.setEnabled(enabled);
                }
            });
        }
    }

Testing

Dynamic methods will not be automatically injected during unit testing, because addons are simply not initialized for this kind of tests. However you can use RmiEnhancer.enhance(metaClassInstance, rmiProviderInstance) where rmiProviderInstance is of type griffon.plugins.rmi.RmiProvider. The contract for this interface looks like this

public interface RmiProvider {
    <R> R withRmi(Map<String, Object> params, Closure<R> closure);
    <R> R withRmi(Map<String, Object> params, CallableWithArgs<R> callable);
}

It's up to you define how these methods need to be implemented for your tests. For example, here's an implementation that never fails regardless of the arguments it receives

class MyRmiProvider implements RmiProvider {
    public <R> R withRmi(Map<String, Object> params, Closure<R> closure) { null }
    public <R> R withRmi(Map<String, Object> params, CallableWithArgs<R> callable) { null }
}

This implementation may be used in the following way

class MyServiceTests extends GriffonUnitTestCase {
    void testSmokeAndMirrors() {
        MyService service = new MyService()
        RmiEnhancer.enhance(service.metaClass, new MyRmiProvider())
        // exercise service methods
    }
}

On the other hand, if the service is annotated with @RmiAware then usage of RmiEnhancer should be avoided at all costs. Simply set rmiProviderInstance on the service instance directly, like so, first the service definition

@griffon.plugins.rmi.RmiAware
class MyService {
    def serviceMethod() { ... }
}

Next is the test

class MyServiceTests extends GriffonUnitTestCase {
    void testSmokeAndMirrors() {
        MyService service = new MyService()
        service.rmiProvider = new MyRmiProvider()
        // exercise service methods
    }
}

Tool Support

DSL Descriptors

This plugin provides DSL descriptors for Intellij IDEA and Eclipse (provided you have the Groovy Eclipse plugin installed). These descriptors are found inside the griffon-rmi-compile-x.y.z.jar, with locations

• dsdl/rmi.dsld
• gdsl/rmi.gdsl

Lombok Support

Rewriting Java AST in a similar fashion to Groovy AST transformations is possible thanks to the lombok plugin.

JavaC

Support for this compiler is provided out-of-the-box by the command line tools. There's no additional configuration required.

Eclipse

Follow the steps found in the Lombok plugin for setting up Eclipse up to number 5.

1. Go to the path where the lombok.jar was copied. This path is either found inside the Eclipse installation directory or in your local settings. Copy the following file from the project's working directory

    $ cp $USER_HOME/.griffon/<version>/projects/<project>/plugins/rmi-<version>/dist/griffon-rmi-compile-<version>.jar .
   

2. Edit the launch script for Eclipse and tweak the boothclasspath entry so that includes the file you just copied

   -Xbootclasspath/a:lombok.jar:lombok-pg-<version>.jar:        griffon-lombok-compile-<version>.jar:griffon-rmi-compile-<version>.jar
   

3. Launch Eclipse once more. Eclipse should be able to provide content assist for Java classes annotated with @RmiAware.

NetBeans

Follow the instructions found in Annotation Processors Support in the NetBeans IDE, Part I: Using Project Lombok. You may need to specify lombok.core.AnnotationProcessor in the list of Annotation Processors.

NetBeans should be able to provide code suggestions on Java classes annotated with @RmiAware.

Intellij IDEA

Follow the steps found in the Lombok plugin for setting up Intellij IDEA up to number 5.

1. Copy griffon-rmi-compile-<version>.jar to the lib directory

    $ pwd
      $USER_HOME/Library/Application Support/IntelliJIdea11/lombok-plugin
    $ cp $USER_HOME/.griffon/<version>/projects/<project>/plugins/rmi-<version>/dist/griffon-rmi-compile-<version>.jar lib
   

2. Launch IntelliJ IDEA once more. Code completion should work now for Java classes annotated with @RmiAware.