Customer Reviews

Executable UML : A Case Study

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

 

 

At the price ([price]on my door step), I expected to find something that put all the xUML pieces together, that was worthy of the author's name. What I did find, though, was a book that provided only the most basic information and examples of the topic, and very incomplete at that. You can find xUML white papers freely available on the web which literally contain 100 times more detail and usable technical information. Not only were many parts of the model missing (specifically key service domains), but key elements such as bridge definitions, use cases, etc., were also missing.

Also, there were many self-serving references to both Leon's company and a future book on xUML. Hopefully, THAT book will contain the useful detail that is missing from this one.

For anyone who is considering purchasing this book, wait for the next one and locate the white papers on the web!

This book was an excellent reference for me as I am still learning the Shlaer/Mellor method, but I really needed to see the Device I/O service domain and the Transport domain to complete my personal project. Hopefully, you can get this out in the next version, as promised, so the model starts "clicking". Well written and documented.

Examples are tough for authors, especially in software. As most software people know, a good example can be as big as the book itself. Often the examples in a book just seem overly simply. Mr. Starr has avoided the problem and has given us an entire book as a complete, worked out example. For those familiar with Executable UML, the efforts to combine UML notation with precise execution rules, this book will be a welcome addition. The Introduction chapter is excellent, tracing many of the pitfalls of current software practice. The choice of an elevator as the project subject matter is also inspired. We all have experience with elevators and think we know how they work. I know I have stood waiting for an elevator to arrive, convinced that I could write a better control program that would result in shorter waits. The model in this book reveals how subtle application problems can be even when the subject is superficially familiar. Most important is the fact that the Executable UML is just that -- executable. The software that comes with the book allows you to do load up the model and try it out. And this is an important point. The intent here is to build a formally executable specification that can then be automatically translated into a running system. The model contains all the semantics to do that. For those of us that have practiced this method professionally, one is struck by the fact that these models have the characteristics of real world models. There are small problems that need to be fixed or reworked Fortunately, Mr. Starr promises additional versions in the future. There are additional domains that also need to be modeled. But in many respects the elevator model is not like what one might see in real world practice. The model descriptions are much better than most I have seen. The quality of the descriptions indicates profound depth of understanding of the problem. It is that depth of understanding which is the key to solving any problem.

The best way to learn a new method is to study an example of how the method is used in real life. This the example!

The introduction to this book goes trough the essentials of the Shlaer-Mellor method from the practising users point of view. About every other page you say to yourself: "Ah - yes - that's how it is...". This since Mr. Starr keeps reminding the reader of some of the odds and ends of the Shlaer-Mellor methodology that might be forgotten while you are busy modelling some small corner of your problem. Very useful.

Even more useful is the rest of the book: - a thoroughly worked out example of how an elevator can be controlled. By choosing a system that we all understand how to use, everything becomes real. The mission of the application and the separation of the service domains become very obvious. Although the models in the elevator example are not fully complete, they give you lots of information and insight.

Especially I like the way Mr. Starr documents his problem space using sketches and figures that clearly explains what's going on and also helps out in the process of finding the right classes and relationships in the model. Studying an example like this helps a lot when building your own models.

Blues immortal Muddy Waters sang:

Well, if the river was whiskey, and I was a divin' duck
Well, I would dive to the bottom, never would I come up.

While many of us have worked on at least one software development project that could "drive one to drink," we borrow here only the singer's craving, which, like that of the duck, is hard to satisfy. With "Executable UML - The Elevator Project," Mr. Starr gives those who drink from the waters of software development a river that flows from its plentiful and pure headwaters, the Shlaer-Mellor Method. So, "... if the river was an Executable UML modeled system..."

But, before we drink, the book is a case study, and, as such; the author is challenged to build a bridge that meets the needs of the communities on both sides of the river. Viewed from one bank, we need some understanding of Executable UML, the author's objectives, the Elevator problem and the logistics of the tool. From the other bank we need to see the Elevator project artifacts just as they would be produced, without the commentary and critique of a case study.

Mr. Starr constructs that bridge in the following manner. On one bank it is anchored with the Introduction and the Guide to Printing Model Diagrams Technical Note. On the opposite bank the bridge is anchored with the Elevator Application domain analysis model diagrams (the Project Technology BridgePoint repository is also included) and two design documents, the Elevator Project Domain Model and Class Model Descriptions. Each span is constructed from a set of five additional technical notes that describe the system requirements and the approach to fulfilling those requirements.

Now, let's swim into the waters of the system's domain model and taste model based design. Dive to the bottom of the executable Elevator Application model, and drink from the completeness of an unambiguous domain specification. Nudge any object and observe its behavior, savoring the flavor of continuity between the data, state and process model elements, and quench our thirst for verification in the Automated Simulation Exercise. Let's sip translated code generated with the BridgePoint model compiler, which is provided on the demonstration disk. Finally, we relish the ability of the process to be repeated, documented, standardized, managed and optimized.

"Executable UML - The Elevator Project" is an excellent study in what is probably software engineering's best practice. Through the book's comprehensive detail, we can increase our understanding of Executable UML and its fundamental, distinguishing characteristics: separation of subject matter, executable analysis models, implementation through translation and adaptability to a comprehensive engineering process. "... Well, I would dive to the bottom, never would I come up."