Customer Reviews

Eclipse Modeling Project: A Domain-Specific Language (DSL) Toolkit

5 star:		(1)
4 star:		(2)
3 star:		(2)
2 star:		(0)
1 star:		(0)

 

 

I need to give a bit of balance to the previous reviews. I basically like the book, but it is a mixed bag and could be a lot better in certain areas. If you expect a start to finish tutorial, you will be disappointed because some background material is presented out of order and you really should read some of his cited references beforehand. For example he discusses the dynamic instance capability in EMF using a mindmap DSL, but doesn't introduce the mindmap DSL itself, until a little later. Also, it will help to have some familiarity with EMF as the coverage in chapter 3 can be followed but is a bit thin and EMF and ecore are purposefully not intended to duplicate the EMF book. The book's usefulness to you will depend upon where you are in MDD and how you want to use MDD and DSLs. The eclipse modeling project is and will undergo rapid change and this book appears somewhat rushed and rough in spots. His second edition when it comes out should be really better organizationally as the Eclipse Modeling projects mature. For example,Chapter 5 on Xtext and TCS is basically a mild description with no example content because that area was not mature at publication. The examples in Chapter 14 on the Xpand template language are in my view a bit skimpy and simplistic for folks wanting to do serious scripting. Similarly if you want to learn QVT, section 13.6 on library operations provides syntax but little to no examples, but the discussion in section 13.5.3 on the shorthand notation used in imperative iterate expressions is really helpful in understanding QVT syntax in the examples presented elsewhere in the book--like chapter 6. Therefore to get the most out of this book, you will be hopping around and be doing multiple reads and passes until it makes sense to you. Don't give up. Do buy the book if you're interested in MDD and DSLs. Gronback has done a nice job developing a mindmap and requirements DSLs that he uses in chapter 6 on model to model transformations. Although you'll need to get comfortable with the QVT and OCL, try running the examples yourself--typing them in to get used to the syntax. His view of MDA at Eclipse in Appendix B is spot on accurate, and I hope OMG takes his advice. If you are interested in graphical model development for your DSL, then Gronback offers strong support for that in Chapter 4 and Chapters 9 through 12. This is no surprise as Gronback is the project lead for GMF. However if you are interested in model to model transformations, QVT, and model to text, and OCL you will be disappointed if you are a beginner requiring step-by-step explanations of the QVT code. Also, don't give up when you reach section 3.2.4 on applying OCL. If you're new to OCL and ecore Eannotations, this will be a rough spot. Finally if you're put off by the price, I'd recommend taking a look at downloading all the OpenArchitectureware tutorials. They are a little stale wrt the later Eclipse Modeling releases, but still helpful and free!

If you use Eclipse, you probably know it as a neat Integrated Development Environment for java, where perhaps you manually write all the java code. This book shows a different take on Eclipse; a much higher level of sophistication. Basically, it shows how to go from a diagram of a set of related classes to autogenerated java code. Developers have put a massive amount of effort into fleshing out this ability; something well documented by the book.

The starting diagram is a domain model. It models a scenario that you are interested in. You might think that an alternative approach is to use Backus Naur Formalism to define the same information. But the book deprecates BNF. Decades-long experience with BNF suggests that it lacks expressive power, and that perhaps the pure text nature of a BNF instantiation is too hard to grasp relationships between objects. Instead, the expression of a domain model in terms of what is roughly [or exactly] a UML diagram gives visuals that people can far more easily absorb.

What the book has is a set of very detailed tutorials. Each walks through a given example model. Fleshing out how to use Eclipse to make the model. Initially, the book's narrative is at a very abstract level, which may hard for some to follow. But the tutorials help give this substance. Each tutorial is non-trivial, and shows how Eclipse has extensive abilities to help you.

In terms of autogeneration, it's not just java code. The book gives examples of autogenerated XML, and in general we see the use of Model to Text transformation templates that spit out text files. One common feature of all of these is the sheer verbosity of the the textual output. Autogeneration saves you labour and reduces the occurrence of simple bugs. Thus Eclipse lets you work as much as possible at a diagrammatic level, which can be more productive.

I am about 1/3 through working the tutorials in this book and really (unfortunately) finding that it is out of date with Eclipse 3.6 (Helios) and the Modeling Project. For example, the tutorials walk you through setting up projects from the Eclipse Amalgam DSL Project and that does not seem to exist anymore (even though the modeling components do in their own sub-projects under the overarching Eclipse Modeling Project). I hope their is a new edition of this book though since the author really does do a great job of explaining the practical aspects of EMF and how to integrate with GMF, XText, OCL, and other modeling project components. I really need some hands on tutorials on these.

I have a number of books in the Eclipse Series (The Eclipse Modeling Framework, The Eclipse Rich Client Platform, etc.) and this one seems to be the pinnacle of the series, describing the fusion of a number of different Eclipse project initiatives into the Eclipse Modeling Project (the EMP). The book is well written, but the topic that it deals with seems to be in a state of 'becoming' rather than 'being'. As a result, a good portion of the book addresses topics and development processes specific to this project that either have already changed, or are in the process of imminent change. This apparently led the author into prescribing a rather rigid workflow for the creation of models and their corresponding tooling, for inclusion as part of an Eclipse application or workbench feature. Personally, I believe that the EMP and GMF are about to hit a point of collapse (in a good sense of the word): a number of portions of the workflow that the author describes really should be refactored to reduce their complexity and streamline the process of model and tool construction and deployment. In its current state, following the EMP workflow is very sensitive to the sequence of operations, and apparently small changes in the sequence can lead to unpredictable results or even failure in the construction of the target. I would definitely buy a revised edition of the book, once the platform that it describes has gone through a couple of revisions and streamlining/stabilization efforts. The author is obviously extremely knowledgeable of the topic, and describes it quite well. He was also able to communicate the fact that the EMP overall is a very important technology in the software engineering landscape, specifically in the areas of model-driven design and development, and the cultivation and continued improvement of the EMP will pay huge dividends in the development of high-quality software. I believe that the sooner the EMP matures and stabilizes, the better! I'm very much looking forward to using it as a primary part of my software engineering toolbox in the future!

ECLIPSE MODELING PROJECT: ACHIEVE BREAKTHROUGH PRODUCTIVITY AND QUALITY WITH MDD AND ECLIPSE-BASED DSLS is for any Java programming library strong in domain-specific languages and software engineering guides. It covers the benefits of a model-based approach and surveys the Eclipse Modeling Project's ability to create new DSLs. An outstanding survey evolves in a technical title recommended for any advanced computer collection.