Browse Books

Reviewer: R. K. Ragade

Let us distinguish software engineering from mere additional programming theory, as it involves human activities in producing reliable and usable software. It also includes large-scale programming activities, management, and quality assurance. In such a diverse and rapidly changing subject area as software engineering, it is desirable for a book to give a summary of checklists and things to do in a practical design environment. It is also desirable for a good textbook to present the basic concepts in software engineering. For a first undergraduate course in software engineering, students who have no previous exposure to the demands of project-oriented software development require templates or patterns of work habits that they can adapt. Software engineering requires project management. There is a need for a book that instills the habit of meticulously integrating the issues of software acceptance with the changes that software design will spawn. Finally, all software engineering texts must stress the documentation of the product as it happens, with annotations for every change. Haste makes waste in any field, including software engineering. Reference works might include detailed compendia of additional findings. Sommerville This is a well-written textbook with a reasonable number of chapters and a reasonable length. Most seniors or graduate students would welcome its size. Each of the twelve chapters is of suitable length and has adequate references for further reading. Sommerville follows the classic waterfall model of life-cycle software development. A third of the book concentrates on design, requirements, and specifications (as it should). Many of the important concepts are highlighted in bold letters. Programming practices and programming language selections are discussed next. Simple examples illustrate steps in the development process. The reviewers wish that the author had also introduced the idea of metrics and complexities, as these could help in the development process. He leaves it to the instructor to supply relevant problems to students. The remaining chapters are rather sparse, given the importance of their subjects in reliable software products. The areas of software testing, modeling, and management are extensive enough that supplemental material would be necessary, according to the instructor's taste, to cover them properly. Yet, this sparse treatment could have been compensated for by better examples of design. Sommerville does give a checklist at the end which could be very useful to a student in a class project. There are few answers to exercises. In an area where there may be no right or wrong answers, or there could be multiple answers, students should feel comfortable in the proliferation of designs. Later in the design process they may be able to choose between the alternatives using several criteria. About 180 references are cited at the end of the book; they provide an adequate sample of the important literature. In summary, the text is light on metrics and models, but is good in providing a usable framework for starting and completing a project from scratch. The book is an outgrowth of a set of classnotes and meets its purpose rather well. It is very useful as a text, given its already wide popularity in many schools (along with Pressman [1]). It is very easy to read this book and to grasp basic concepts. The second edition has incorporated a number of changes. Although the treatment in the book is still qualitative, it is comprehensive and follows the waterfall model of software development fairly closely. The author treats the subject enthusiastically with up-to-date examples. Lamb At 298 pages, this book is long for its intended purpose. True, it has a number of sample templates at the end for major software engineering project activities, making it useful to a project-oriented course, but as a textbook it is heavy on documentation, and apart from documentation and algebraic specification treats most topics either sparsely or not at all (this latter topic is not addressed in most other software engineering texts). The major point in favor of this book is that it could be a companion to some other book, such as Sommerville's or Pressman's [1] texts. It includes checklists and templates. In a textbook, there is a need for explicit definition of concepts. The author does give examples at the end of each chapter, in the form of project ideas. As a text it is weak in testing, quality assurance, reliability and maintainability, and software economics. These topics also ought to be covered in a software engineering course. The 20 chapters and 10 appendices could be compressed into a fewer number for useful reference. The book gives a small number of references (about 43), but omits any reference to the works of Pressman or Sommerville. Pressman In contrast, this book can also serve as a companion reference text and it does remarkably well in 212 pages and 10 chapters. Although there are no examples, the author does provide a well-organized collection of over 90 good references. In addition, numerous checklists, a self-assessment questionnaire, and practical tips based on years of hands-on experience make this book a delightful read. As in the case of Lamb's book, Pressman begins with a focus on planning and managing change. But greater emphasis has been placed here on the practice of software engineering management and of issues in software engineering, as well as on software quality assurance. By incorporating customer training into software quality assurance and acceptance, Pressman has consolidated ideas that are increasingly popular with companies marketing large-scale software, particularly fourth-generation and CAD/CAM simulation modeling and expert systems. Technology transfer should be an important concept in the software design. In summary, the book serves its intended purpose as a reference and companion guide excellently. It is very useful and very readable, with ideas in bold print. This book or its equivalent should be required reading for any chief programmer in any software development team. Another book that is comparable is Birrell and Ould [2]. Conte, Dunsmore and Shen The main emphasis of this book is on the role of metrics and models in software development, particularly relating to quantification at the program development level. At about 400 pages it is just right, neither too long nor too short. There are 8 chapters, with typically 10 to 15 exercises per chapter. Some of these relate to program analysis, while others relate to data analysis of specific program collections. The exercises given are not adequate for use in a graduate class; the instructor must substantially augment these exercises. As with many software engineering classes, term projects help in forcing students to apply these concepts to the programs in the project. The book has perhaps the best collection of references on software metrics and models available at the time of publication. Its primary purpose is more as a reference or secondary text. The goal of the authors was to bridge the gap between high-level and working references in software engineering education, where practitioners could have reasonable access to metrics and models that they could experiment with. The empirical flavor of software engineering has been well established, with sufficient attention paid to the issues of experimental designs in software development and testing. In this effort the authors have done a service to the community by stressing the need to develop quality software with metrics and careful experimentation and testing. The authors present the material in a logical order, progressing from the issues of `program metrics' to `software metrics' and later to software costs and software management issues. They also make the general software engineer aware of the importance of human factors in software development, at both the cognitive and the social interaction levels. Pfleeger This book is at the level of an undergraduate text in software engineering, and is more suited for classes in software technology. It gives a general understanding of software engineering practice for the information systems or MIS person. While many of the basic concepts of the software development cycle have been simply and lucidly explained to give an introductory flavor, they leave the reader with a large number of buzzwords, such as Petri nets, which are, of course, not explained in any way that the student can immediately use. Each chapter has an inadequate number of examples for an undergraduate text, with none too specific. Students at this level need less uncertainty and more practice with concepts. The chapters on design are weak. The chapters on program testing, maintenance, and management have been hurried through. Very little workable knowledge for doing any reasonable software cost estimate has been imparted to the student; this knowledge is crucial for obtaining quality and affordable software. The usefulness of this book is more as a light tutorial, as it is readable with moderately large print and nice diagrams and sketches. It may be a useful text to include in those curricula other than software engineering or computer science where computer software development is equally crucial. Fairley This book gives a reasonable amount of attention to a balance of various topics in software engineering, ranging from design to cost estimation and management. The author brings a fresh viewpoint based on his vast experience with software development and software engineering. He emphasizes some of the more current trends in software engineering, including empirical considerations from large-scale software development. Beginning with planning and issues of software cost estimation, the author describes the main variations of the software development cycle, including the fundamental principles of structured design and formal methods of determining specifications. The author also stresses the methodologies for software development and software project management. The interplay of the two is in fact crucial to understanding the production of large-scale software, where errors can creep into a design from ambiguities and uncertainties of design specifications and implementations. The author leads the reader into issues of software maintenance and software quality assurance. He discusses some aspects of software metrics, though not as thoroughly as Conte et al. do in their book. A major feature of this book is its emphasis on group and term projects for students. In fact, the author considers these to be essential as a means for the instructor to explain basic concepts of software engineering. Without a collection of well-thought-out projects, the exercises provided can be at best a weak approximation. The author has also provided suggested formats for documenting term projects. It is well known that good documentation is half the battle in getting a software product to be accepted and well maintained. Students should often be encouraged to pause and write about what they have implemented, before they forget the significance of major and minor changes during software evolution. When I teach software engineering I place a major emphasis on students turning in a user's manual, an owner's manual (which is really an elaboration of the user's manual, along with maintenance information), and a software analysis document that includes estimates of cost, reliability, complexity, and other metrics. As the title of the book suggests, each section addresses one of the major concepts in software engineering. This makes it easy for a graduate student to read just what is sufficient for carrying on with term projects. Any reader who needs elaboration on concepts or methods would do well to turn to the numerous references provided by the author, who also recommends some well-known journals and periodicals in this area for companion reading. The book is about 350 pages long and has good print. Each chapter has a small number of choice references that would not overwhelm a graduate student. In some other books, all references are given at the very end. Fairley has chosen to be specific for each chapter. Comparison The six books in this review have specific audiences in mind. This explains the variation in emphasis and quality. Table 1 gives some facts which could aid a reader in evaluating the format and layout of material, and subjective comparisons of each book are summarized in Table 2. The reviewers believe that all these books have sought to fill specific niches. The choice of a textbook is heavily influenced by the didactic style of an instructor. If one of these books is chosen as a textbook for a graduate or an undergraduate class, the others could easily fill the roles of companion references.