Browse Books

Reviewer: Pavol Navrat

Sommerville is an active researcher, educator, and organizer in the field of software engineering. This book consists of a preface, 31 chapters, an appendix, and indexes. It is structured in five parts. After two introductory chapters covering human factors in software engineering, the first part is devoted to software specification. It includes chapters on software requirements definition, system modeling, requirements specification, software prototyping, formal specification, algebraic specification, and model-based specification. All the notions are introduced and explained lucidly. The difference between prototyping and exploratory programming is discussed. Z schemas are introduced for model-based specification. The second part covers software design. The author includes chapters on object-oriented design, function-oriented design, real-time systems design, and user interface design. Sommerville considers the two principal design strategies, namely functional and object-oriented design, to be complementary rather then opposed. Each may be applicable at different stages in the design process. Programming techniques and tools are discussed in Part<__?__Pub Fmt hardspace>3. Chapters on programming for reliability, software reuse, computer-aided software engineering, and software development environments show that the author concentrates on some advanced programming issues, and of course tool support for the software development process. On the other hand, programming itself is not covered; this omission is common in software engineering textbooks and monographs. Part 4 is devoted to software validation. The author has chosen verification and validation, software reliability, software safety, defect testing, testing and debugging tools, and static verification as the main topics here. The last part covers software management. It includes chapters on project planning and scheduling, software cost estimation, software maintenance, configuration management, documentation, and software quality assurance. A typical book on software engineering covers a broad range of topics. As a consequence (to keep the size reasonable), it is not deep. The essence of almost all important notions, methods, techniques, and so on is introduced. Usually, as the authors give their assessments of the methods, they evaluate the important trends in the field. Rather than describing the basic methods in sufficient depth to serve as a genuine source, they give only a sketchy introduction, and the stress is on their properties. When such a book is used as a textbook, this approach may cause difficulty for the student. In contrast to the case of software engineering, a typical book on any well-established field of engineering, such as electrical engineering, covers basic methods of the field in enough depth to serve as a source of knowledge and a reference for both students and practitioners. The stress in such books is on practical knowledge. The meta-level knowledge, which includes evaluation of methods, is implicit: the inclusion of a method in the book is the indication of its importance. The situation in software engineering books clearly has its roots in the immaturity of the field, especially when compared to more traditional fields of engineering. The ongoing discussion of the differences and boundaries between computer science and software engineering proves that even these fundamental questions are not yet clarified, or that the proposed answers are not yet universally accepted. What can we realistically expect from a book introducing a complex field of engineering that is still in its infancy__?__ Sommerville's book is an encouraging attempt. While it obviously cannot overcome the difficulties inherent in the present state of the art, it does what can be done in an outstanding way. The book is broad rather than deep. As long as we do not have one or two standard, universally appli<__?__Pub Fmt hyphen-point>cable specification, design, implementation, and maintenance methods, this approach seems inevitable. The author has done his best to minimize its shortcomings. He covers almost all the important issues, not restricting his scope to management or to systems analysis and design, as several other books do. His deep insight into the matter is presented in a condensed form and in a readable way. The form of the book is another of its good points. Each part starts with a short statement summarizing its objectives and a list of chapters. Similarly, each chapter starts by stating its objectives and contents. At the end of each chapter, the key points are summarized, further reading is indicated, and exercises are provided. The author and subject indexes are adequate. The list of references is comprehensive. The book is aimed at both students and practitioners of software engineering. For both, it is an excellent source of knowledge. Practitioners will find relevant chapters to update their knowledge or to get a clear explanation of particular topics. Despite being too long for most one-semester courses, the book can be used as a textbook for either an introductory course or some advanced courses, such as Requirements Engineering, Reliable Systems, and Project Management. The book has benefitted from gradual improvements in newer versions. Among the changes from the third edition<__?__Pub Fmt hardspace>[<__?__Pub Caret1>1], the inclusion of new chapters on real-time systems and software safety, and more detail about prototyping and software reliability, are worth mentioning. The material on programming and abstract data type implementation has been shortened. The book can be recommended as useful for both the intended purposes. When it is used as a textbook, however, the assumption of previous familiarity with programming and modern computer systems must be taken into account.