At ESE, I had a nice chat with Chris (actually, I had a lot of nice chats with a lot of nice people - it was quite a challenge to attend any of the sessions), who told me that he was looking into template engines. I leave it up to you to make any assumptions on why he's interested in template engines. I happen to know at least three template engines: Velocity (which I had the joy of using in my active days at AndroMDA.org), the template, erhm, ... mechanism being used in PDE (I once fixed a bug which had been caused by this engine) and - you might have guessed it - Xpand (which I am a committer on).

So, I gave a short demo of Xtext and Xpand to show Chris how easy it is to create (model-aware) code generation templates. Xpand comes with a nice editor which makes template editing a joy - it features code highlighting, hyperlink navigation, model-awareness and other editor goodness.

Instead of letting only Chris know how that works, here's a short guide on how to write your own Xpand template and use it in a wizard. To make the example more realistic, I chose to implement a wizard that creates an Ant build.xml file for a simple project (something which is missing in Eclipse, AFAIK). So, here we go.

Prerequisites:

• Eclipse 3.5M3 (3.4 might also do)
• EMF 2.5.0M3
• EMF Compare 0.8.1
• MWE 0.7.0M3
• Xpand 0.7.0M3

How to do it

1. Download and install the prerequisites.
2. Create a new plug-in project named "de.peterfriese.antwizard"
3. Add a New File Wizard to the project. I used the Custom plug-in Wizard to get me started.
4. Brush up the wizard page and add some fields for project name, source folder and binary folder.
5. We want to transfer the values entered in this wizard into our template, so we need to create an Ecore model that describes our data model.
   

6. Add org.eclipse.emf and org.eclipse.emf.edit to the dependencies of you plug-in.
7. Create a genmodel and let EMF generate the model code for you.
8. Add code to your wizard that transfers the values from the input fields to your model. One might consider using databinding for doing this, I used a straight forward read'n'write approach for the time being:

   public boolean performFinish() {
       final String containerName = page.getContainerName();
       final String srcDirName = page.getSrcDirName();
       final String binDirName = page.getBinDirName();
       // ...
   }
   private BuildSpecification createModel(IProject project, String srcDirName,
       String binDirName) {
       BuildSpecification buildSpecification = BuildspecificationFactory.eINSTANCE.createBuildSpecification();
       Project projectSpecification = BuildspecificationFactory.eINSTANCE.createProject();
       projectSpecification.setName(project.getName());
       projectSpecification.setBinaryFolder(binDirName);
       projectSpecification.setSourceFolder(srcDirName);
       buildSpecification.setProject(projectSpecification);
       return buildSpecification;
   }

9. Create an Xpand template de.peterfriese.antwizard/src/template/BuildTemplate.xpt:
   

private void generate(BuildSpecification buildSpec, IProgressMonitor monitor) throws CoreException {

    // get project root folder as absolute file system path
    IWorkspaceRoot root = ResourcesPlugin.getWorkspace().getRoot();
    IResource resource = root.findMember(new Path(buildSpec.getProject().getName()));
    String containerName = resource.getLocation().toPortableString();

    // configure outlets
    OutputImpl output = new OutputImpl();
    Outlet outlet = new Outlet(containerName);
    outlet.setOverwrite(true);
    output.addOutlet(outlet);

    // create execution context
    Map globalVarsMap = new HashMap();
    XpandExecutionContextImpl execCtx = new XpandExecutionContextImpl(output, null, globalVarsMap, null, null);
    EmfRegistryMetaModel metamodel = new EmfRegistryMetaModel() {
        @Override
        protected EPackage[] allPackages() {
            return new EPackage[] { BuildspecificationPackage.eINSTANCE, EcorePackage.eINSTANCE };
        }
    };
    execCtx.registerMetaModel(metamodel);

    // generate
    XpandFacade facade = XpandFacade.create(execCtx);
    String templatePath = "template::BuildTemplate::main";
    facade.evaluate(templatePath, buildSpec);

    // refresh the project to get external updates:
    resource.refreshLocal(IResource.DEPTH_INFINITE, monitor);
}

That's it!

Taking it for a spin

1. Launch a runtime workbench by selecting de.peterfriese.antwizard/META-INF/MANIFEST.MF and invoking Run As -> Eclipse Application
2. Create a new Java project in the runtime workspace.
3. Invoke your wizard by selecting File -> New -> Other... -> Ant -> Ant build file from existing project
4. Enter the required information (project name, source folder, binariy folder)
5. After clicking on finish, you should get a fresh Ant file for your project:
   

You can download the project here. Feedback and comments are welcome!

Example

importMetamodel "http://www.example.org/my/bsl" as mybsl;
A:
 (imports+=Import)*
 (bs+=BRef)*;
Import:
"import" file=URI;
BRef:
"a" name=ID "b" b=[mybsl::B];

B:
"b" name=ID;

Install an xtext language

package org.example.bsl;
...
public class EPackageProxy implements EPackage {
  static {
    ResourceLoader cl = ResourceLoaderFactory.createResourceLoader();
    try {
      ResourceLoaderFactory
        .setCurrentThreadResourceLoader(new ResourceLoaderImpl(
           EPackageProxy.class.getClassLoader()));
      EPackage.Registry.INSTANCE.remove("http://www.example.org/my/bsl");
      MetaModelRegistration.register();
    } finally {
      ResourceLoaderFactory.setCurrentThreadResourceLoader(cl);
    }
  }

  public static final EPackage eINSTANCE = MetaModelRegistration
       .getEPackage();

  public EClassifier getEClassifier(String name) {
    throw new UnsupportedOperationException("Method not implemented");
  }

  // remaining methods of EPackage
  ...

  <extension point="org.eclipse.emf.ecore.generated_package">
     <package class="org.example.bsl.EPackageProxy" uri="http://www.example.org/my/bsl">
     </package>
  </extension>

Referencing an installed xtext language by nsURI

package org.example.asl;
...
public class RegisterBslHelper {

   public RegisterBslHelper() {
     EPackage package1 = MetaModelRegistration.getEPackage();
     package1.eResource().setURI(URI.createURI(package1.getNsURI()));
   }
}

<workflow>
  <property file="'generate.properties'/">
  <bean class="org.example.asl.RegisterBslHelper">
  <component file="'org/openarchitectureware/xtext/Generator.oaw'" inheritall="'true'/">
  </component>
</workflow>

Jeffrey Overbey talked about building refactoring tools using the LTK and his research work "Ludwig".

I guess you all agree with me that refactoring support is crucial for any programing language, as you need to restructure your code rom time to time.

Although Eclipse LTK offers a decent API and UI to realize refactoring tools, all the heavy lifting (parsing the text in question, analyzing the AST, creating rewrite rules) is left to you.

Jeff's research tool Ludwig can help to derive refactoring support from a BNF-style grammar very easily. Ludwig essentially derives a lexer/parser from the grammar and provides an interface to the AST which gets constructed by the parser. Which gives you the chance to walk / visit the AST and (partially) rewrite the AST.

I really enjoyed Jeff's presentation style. He used a set of slides and pre-recorded screenvideos to drive his talk. I think this is great idea, as it basically eliminates any problems you might run into if you do a live demo. Plus, it gives the presenter the chance to face the audience while demoing, instead of mumbling into the screen

On November 10th, the Hamburg Eclipse DemoCamp took place in the stylish EAST Hotel.

We had 45 reservations and almost all of them made it to the Camp. I always get a little bit nervous 15 minutes before the event starts if no more than 10 people have shown up so far, but I guess that's alright.

Martin and I welcomed the crowd on behalf of our sponsors (Eclipse Foundation, froglogic, it-agile and itemis):

After that, Harald Wellmann of Harman Becker told us that the world is a disc. Well, at least he and his company try to make it a disc again - Harald leads a team that develops a so called "map compiler". A map compiler takes map source data and condenses that data by extracting only the relevant parts of it. As you may guess this is a long-running process which can hugely benefit from parallelization. Harald and his team use OSGi to modularize their software and make sure they use computing resources efficiently. One thing worth noting is that OSGi is even being used in car entertainment systems: Harald told us about one entertainment system which makes use of OSGi to act as an intermediate / glue layer between the UI (written in Java) and the core (written in C++).

Gerd Wütherich (Independent) continued where Harald stopped and showed us how to use Spring Dynamic Modules and OSGi in his excellent talk - interspersed with some neat demos:

After those talks, we took a break to grab some refreshments and take the chance to get in touch with the other attending Eclipse enthusiasts. I had the impression that everybody had a good time discussing all things Eclipse - in fact Martin and I had to interrupt a lot of lively discussions for the second run of talks.

In the first talk after the break, Miguel Garcia and Rakesh Prithiviraj (both Technical University of Hamburg-Harburg) gave us an update of their research on how to integrate LINQ (Language Integrated Queries) in Java:

The final demo was deliverd by Stephan Herrmann who showed us Object Teams / Equinox, an amazing piece of software that can be used to re-use existing Eclipse plug-ins in an aspect-oriented way. To get an idea of how powerful this approach is, have a look at the following screenshot - this is the JDT, but enhanced by Object Teams in order to support their very own syntax extensions for Java:

The feedback we received from the attendees was great - some people even sent emails thanking us for organizing the event, so I guess the DemoCamp can be considered a success!

• Scala - (It's an introduction to a programming language called Scala. Most of the slides contain code snippets only, so it may not be that usable without explanation.)

• Vermisste Sprachfeatures in Java (it's in german and is about some IMHO important language features Java lacks.)

It's that time of year again - demo time!

Being run for the third time, demo camps can be considered an integral part of the Eclipse calender of events. Attending a DemoCamp is great. If you haven't yet been to any Eclipse DemoCamp, let me briefly summarize why I think you should consider going to one:

• see cool technology in action (well, that's the whole point of a DemoCamp, isn't it)
• get to know Eclipse-minded people from your area
• have some frosty beverages and fingerfood (well, that actually depends on how the camp is organized)
• see some exciting buildings from inside (again, depends on how the camp is organized)

Oh, did I mention that Eclipse DemoCamps are free? Thanks to the Eclipse Foundation and many sponsoring member companies, you get to enjoy all those nice things for absolutely free.

Here are some of my favorite demos:

• Jingwen Owen Ou on WikiText for Mylyn
• Jochen Krause on Eclipse 4.0
• Harald Wellmann on Geodata Processing for Car Navigation Systems
• Florian Fieber and Max Bureck on From Model to Code - A Comparison of M2M and M2T Technologies in Eclipse Modeling
• Dave Savage on Newton - a distributed OSGi application runtime
• Jens von Pilgrim on GEF goes 3D: GEF3D
• Eike Steppe on CDO Model Repository
• Boris Bokowski and Eric Moffatt on the e4 Programming Model and e4 Modeled UI, respectively
• Martin Taal on EMF/Teneo

By the way, if we were to award the price of the most indefatigable DemoCamp presenter, it'd go to Wassim Melhem. You can meet him at no less than three locations: Iasi (Romania), Warszawa (Poland), Poznan (Poland).

Hope I could convince you to attend one of the many DemoCamps. I'll be at the DemoCamps in Hamburg and Leipzig, so see you there!

From October 15 to October 17, MDSD Today 2008 took place at the Nordakademie in Elmshorn near Hamburg, Germany. The conference, which had been organized by Frank Zimmermann (Nordakademie), Simon Zambrovski and yours truly, was intended to bring together practioners, researchers and people from both industry and business.

We were fortunate enough to aquire a number of well-known speakers, most notably Axel Uhl (SAP) and Ed Merks (of EMF fame).

On the first day, Axel and Ed delivered their excellent keynotes on the current state of affairs in modeling.

Ed commented on a number of misconceptions he has been confronted with during the last few years such as "modeling is too complicated", "modeling will just get into my way" or "modeling will limit my creativity". The bottom line of his talk was that although you might not realize it, models drive most software (just think of all the data models you're dealing with in business applications). However, using source code as a representation for your models might not always be the best solution - to gain both a better overview of your software and be more efficient with respect to software development, you should think about raising the level of abstraction, e.g. by using domain specific modeling tools that let you focus on the structure of your model instead of having to deal with (basically irrelevant) syntactical details of the target language. If working at a higher level of abstraction makes things more complicated for you, you're doing something wrong.

Axel picked up this idea in his keynote on "Current Challenges for Industrial Software Development Tools and Languages" stating that any computer program can be conceived as a model. Since models are independent of their visual representation, they can be represented in various different forms, the most common being text, trees and diagrams of any kind. The question, however, whether to choose text or the abstract model as the primary artifact for storing. A textual representation clearly has advantages with respect to diffing and merging and can be handled with commonly used tools such as CVS nd plain text editors. Using the abstract model as the primary artifact allows for overlapping partial views. Since this model can persisted in some kind of repository, access to the model very much feels like working with a database. In particular, there's no need to parse any text back and forth between the model and an editor. The decision whether to store models as text or in a repository largely depends on boundary decisions, like the availability of the repository technology, the robustness of the mapping between the textual representation and the model, language characteristics, the number of developers who need to work on the model in parallel, the size of the software system built with the model and the life cycle of the software. Axel concluded that on one hand, modeling is not that much different from coding and that, on the other hand, DSLs are becoming increasingly important due to the complexity of UML.

One of the nice thing about conferences is you get to talk to people who are interested in the same topics as you are. MDSD Today has been no different - during the breaks you could see people involved in all kinds of discussions.

Later that day, Stefan Reichert and Birger Garbe of Lufthansa Systems shared their experiences regarding model driven software projects with us. The topic of their talk was "Model Driven Software Development in Business Projects: Chance or Risk?" Their anwser to this question was that if've got the right tools, MDSD is a chance and can help you to build better software.

Thomas Stahl of b+m concluded the day with his talk on how MDSD, BMP and SOA fit together:

The second and third day of the conference were crammed with hands-on tutorials on MDSD tools like EMF, Xtext, Xpand and GMF. Quite a lot of people attended the tutorials as you can see from the pictures.

Arno had to overcome the hardship of a broken beamer, so he had to use arms and legs to explain:

Overall, the conference was a real success and I am looking forward to seeing you all again at MDSD Today 2009!

• Sharing an editing domain among several editors
  
• Make GMF's development process easier, e.g. by bypassing unreliable reconcilers and provide a very simple UI to the code generator.
• Additional layouts, e.g. make labels fit into ellipses.
• How to implement non- or semi-canonical diagrams.

    /**
  * @param args
  */
 public static void main(String[] args) {
  System.out.println(S(/* 
     Wow, we finally have
      multiline strings in
       Java! HOOO!
          */));
 }

                                 Wow, we finally have
                                 multiline strings in
                                 Java! HOOO!

   public static String S() {
       StackTraceElement element = new RuntimeException().getStackTrace()[1];
       String name = element.getClassName().replace('.', '/') + ".java";
       InputStream in = getClassLoader().getResourceAsStream(name);
       String s = convertStreamToString(in, element.getLineNumber());
       return s.substring(s.indexOf("/*")+2, s.indexOf("*/"));
   }