Customer Reviews

Building Reliable Component-Based Software Systems

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

 

 

Oh no, another book with a misleading title...

It should have been called "Building ***snip*** Component-Based Software Systems". See the difference ? The only reason I chose to read it was that magic "Reliable" word, and now what do we have ?

So, magic mirror, please tell us who should read this book ?

Quote:
"Experienced developers will find useful technical details ... while inexperienced developers can learn about the principles ... the book is [also] aproppriate as a course book ... for graduate ... or undergraduate students."

I for one am an experienced developer, now where are my technical details ?

Before I continue, please keep in mind that my opinion is biased by the word "Reliable" in the title. The book hardly says a word about reliability, for God's sake.

The book is a compilation of assorted chapters written by different European researchers. Researchers more than practioners that is. Basically the book hardly says anything you wouldn't know if you have ever put your hands on OOP.

First chapters come to components from different perspective each, but most remain introductory. Component is this, interface is that, environment is also concerned... Components are good (but too complex to say anything else)...

Many of the non-trivial conclusions are highly controversial. Example, chapter 3, "Architecting Component-Based Systems". The authors take 3 "architectures", namely "Pipes and filters", "Blackboard" and "Object-oriented", give each a half a page discussion, and come to the conclusion that out of the three the "Pipes and filters" is the most secure (!). Excuse me, but this is absurd.

Various discussions, ex. semantical integrity (limited with design-by-contract), roles, failure injections, product lines were reasonably interesting.

Chapters 11, 12 touch the Koala platform developed at Philips and used internally for their TVs and stuff. Ch. 12 was interesting to read, and is the only one that comes close to the promised "technical details". It's counterpart, ch. 11 is burdened with controversial conclusions, ex. "For example, the use of Windows results in applications based on message loops, which may not be the best architecture in all cases" or "A company that sells a software package can only survive if it has many customers".

The rest of the chapters is dedicated to using components in real-time systems and describing several case-studies. What do they say ? Using components is real-time is more complex than ever... It's difficult to introduce common approaches where machine power is limited... Real-time is special... Well, what do you know.

I won't discuss case studies which take last several chapters. They are various, some too aerial, some not, but they certainly say little new. Emulating a RTOS on top of Windows NT by means of a hi-res timer and cooperative user-level multithreading ? Tag-your-stuff-and-put-it-in-Active-Directory kind of system ? COM-like something which is not quite COM ? It's not RELIABLE.

The writing style is far less than perfect. Look, I understand that the book is written by a bunch of European professors, moreover, I'm not a native English speaker myself, but !

How shall a reader cope with something like: "There is that system, it has two parts to it - BOM and BOF. BOM does this, BOF does that." ? Oh, BOM it is, huh ? Thank you very much, it's all clear to me now. I mean, if you describe a system which is obviously transparent to you, why not make it easy for reader to understand ?

Also fun is repeating same things multiple times within a chapter. Like "Foo is bar, but biz is not baz. ... 2 pages ... Foo is bar, but biz is not baz." First time you encounter this, it's like "wait a minute, I've just seen exactly same thing... flip... flip... oh... what a heck ?". Again, I agree that one big book is better than one small book, but it's somehow disturbing.

I also didn't understand having a UML diagram with Name being a separate entity linked to all other entities on the stage. I always thought that Name should be an attribute.

And on and on it goes...

Therefore 3 out of 5, may be less, may be more for students.

This book is based on papers written by an array of experts in all component-based sub disciplines, and cover the full spectrum of topics. Organization of the book is seven areas, each of which contains relevant papers that start at the general level and segue to specifics. Readers new to component-based software engineering can get the basics from the general-level material, then drill down into details in subsequent chapters in each of the areas presented in the book.

While I found the material uniformly excellent from the perspectives of consistent editing, technical accuracy and best practices, the strongest point is the editors have selected material that manages to encompass all facets of the discipline. This book is all meat with no fluff.

I like the sequence of topic areas, which start with an overview of components, definitions and specifications, to architecture, to development. In addition, subsequent topic areas cover using components, applying component-based software engineering to product line development, and real-time components. When this book was written real-time components were more in the domain of specialized device manufacturers, but this material is also applicable to certain types of web services. In fact, given the growing trend towards web services the developers in that domain can learn much from the work of the component-based software engineering community because there is a lot of overlap between the two at many levels. More importantly, the sub discipline of reuse, covered in this book, is conspicuously missing from many web services development environments. The knowledge in that area alone would make this book a worthwhile investment.

This book is one of two that I believe capture the major elements of the body of knowledge of this discipline (the other is "Component Based Software Engineering: Putting the Pieces Together" ISBN 0201704854). I also strongly recommend "Testing and Quality Assurance for Component-Based Software" ISBN 1580534805 as a supplement to this book if you are working in the component-based software engineering environment and/or are working with web services and wish to tap into the mature and proven best practices and body of knowledge from the CBSE discipline.