What is generic programming about? Every programming paradigm is based on a specific kind of abstraction that distinguishes it from other paradigms. For instance, the abstraction supported by most contemporary programming languages is that of a subroutine (or procedure or function): You can use a subroutine without knowing anything about its implementation. Another abstraction is that of abstract data types: You can manipulate and even create values without depending on the actual representation of the data; only the interface is important. Object-oriented programming adds another layer of abstraction by supporting hierarchies of polymorphic data types: You can refer to a value and manipulate it without needing to specify its exact type. In this way you can write a single function that will operate on a number of types within an inheritance hierarchy.

 The new abstraction that characterizes generic programming is that of a set of requirements on data types. You can write general and reusable functions (more precisely: function templates) that are independent of the data types they operate on. It is a technique comparable to functions that operate on polymorphic types in an inheritance hierarchy. The notion of type requirements comes into play with the template language feature in C++. If you've been programming with templates, you will have noticed that each time you define a template that takes a type argument you make assumptions about the operations that the type must provide. Most often, we do not formally specify the set of requirements to such a template type argument, because C++ does not have any specific language features to express such requirements on data types. And here the is book most fascinating: It is the first publication we've seen that recognizes the need for a systematic approach to template programming. A formalized specification of type requirements?Austern calls them "concepts"?is important for intense use of templates.

 The STL is an example of the generic programming paradigm. It is a collection of predefined class and function templates, but it is also a collection of "concepts" that specify how the STL components collaborate and how the STL can be extended. Without doubt, this book is unique because of the approach it takes in describing the STL in terms of "concepts." The plain use of STL containers and algorithms is described elsewhere; all STL books published so far explain what the predefined STL components have to offer. None of them provides much information about the systematic specification of type requirements behind the STL. This book takes the reader one step further. It describes all predefined components in the STL, but it explains them as manifestations of the abstract concepts. If you just want to use the predefined STL components, this is not the optimal book for you. If you strive for a thorough understanding of the STL and want to use it intensively and intend to extend it, you'll find all necessary information here.

 In sum: the reader will learn a lot about the underlying concepts of the STL and will close the book having gained valuable new insights into the nature of template programming in general. 

Angelika Langer develops and teaches classes on advanced C++, STL, multithreading, internationalization, and Java. She has served on the ANSI/ISO C++ Committee since 1993. Klaus Kreft is a software architect and consultant with more than a decade of experience in industrial software development. He works for Siemens Business Services in Germany. Langer and Kreft are working on a book about standard C++ iostreams and locales and are columnists for C++ Report .