In plain English: The focus is on system families rather than one-of-a-kind systems. Instead of building a single software system as a solution to a certain problem, generative programming is about designing a system from which configurable solutions can be generated based on an assembly plan or a configuration specification. The problem with current software development processes is that we usually end up with a concrete software system, but don't know how we got there. Most of the design knowledge is lost, which renders software maintenance, evolution, and reuse rather difficult and costly. In generative programming, the idea is capturing as much of the production knowledge as possible. Hence GP is about designing and implementing a generative domain model (preparing for reuse) and designing and implementing another model for using the generative model to produce concrete systems (the actual reuse). 

None of that sounds overly familiar, doesn't it? Well, it can't. GP is a new paradigm and this book (or at least part of it) is the result of a research project. Hardly anything of what the authors talk about is common practice today, but it might be in the future. In any case, Generative Programming: Methods, Tools, and Applications is thought-provoking and interesting reading for those who do not mind being confronted with new ideas and being overwhelmed with detailed studies of all kinds of aspects related to GP. 

Let's see what the authors cover: 

• Analysis and Design. Because GP is about designing and implementing a generative domain model, the first part of the book is devoted to analysis and design methods, including domain analysis (determining the scope of the software system families of interest) and how it relates to object-oriented analysis and design, and feature modeling (investigating commonalities and variation among family members).
• Implementation. The second and major part is about implementing the system family model. Although the authors include a discussion of implementation techniques using all conceivable programming languages, they clearly emphasize C++ programming techniques. The chapters on generic programming, component-oriented template-based C++ programming techniques, and static metaprogramming in C++ are among the best coverage of modern and advanced template programming techniques that we've seen so far. Indeed, the chapter on metaprogramming impressively demonstrates the potential but also the limitations of template metaprogramming. 

 

 
 
 
 
 
 
 

In stark contrast, there is a chapter about Intentional Programming, an extendible programming environment developed at Microsoft that allows you to load extension libraries to add domain-specific programming language extensions as needed for a given application. Extension libraries can extend the editor, debugger, etc., and provide an ideal basis for metaprogramming support.

• Application Examples. The third part provides concrete examples of applied generative programming?all using C++ as an implementation language, heavily using templates. The examples guide you through all stages of the generative programming process from domain analysis and feature modeling over domain design to implementing components, design of domain-specific languages, and implementation of generators. It's all done in C++, and it looks weird, but the third example of a matrix computation library comes close to industrial-strength numeric libraries that exist in practice. 

Angelika Langer develops and teaches classes on advanced C++, STL, multithreading, internationalization, and Java. She served on the ANSI/ISO C++ Committee from 1993 to 1998. Klaus Kreft is a software architect and consultant with 15+ years of experience in industrial software development. He works for Siemens Business Services in Germany. Langer and Kreft are authors of "Standard C++ IOStreams and Locales: Advanced Programmer's Guide and Reference" (Addison-Wesley, 1999) and are columnists for C++ Report magazine.