The authors define their goal as providing insights into the design decisions and philosophies of the OS/2 operating system, and they claim to discuss the motivation, architecture, and realization of OS/2. The reader will find a small dose of motivation, an adequate description of the architecture, and heavy emphasis on implementation. Insights into the design decisions and philosophies are rarely explicit, leaving much to the reader’s imagination. Despite this, the authors provide a wealth of information about both the 16-bit and, more important, the 32-bit versions of OS/2. They are particularly careful in presenting the characteristics of the application programming interface (API) for the component parts of the operating system for both versions and in presenting the material in a way that is consistent with the abstractions found in traditional texts while being explicit enough to describe a real product. The book draws on Deitel’s academic and text writing experience and on the real-world design and implementation of OS/2 by Kogan, the chief architect of OS/2 2.0. The authors also clearly describe the roots of the constructs found in OS/2, whether they come from the legacy of DOS and the IBM/Microsoft collaboration or from the UNIX community. One of the more interesting, albeit short, discussions concerns Windows 3.0 and its relationship with OS/2. The authors treat this potentially volatile subject carefully with a relatively factual presentation of the strengths and weaknesses of the Windows implementation.

The first three chapters introduce the history and evolution of DOS and OS/2, hardware chip technology (8088/8086 and 80x86, including a glimpse at future RISC considerations for the i860 and IBM POWER architectures), and the hardware architectures of the IBM line of personal computers. Chapter 4 is an overview of the architecture of the  16- and  32-bit versions of OS/2, while chapters 5 through 10 describe the detailed implementation of the major components of both versions. Their contents include multitasking (chapter 5); memory management (chapter 6); interprocess communications (chapter 7) via shared memory, semaphores, signals, pipes, and queues; input/output (chapter 8), including file systems and device drivers; graphical user interfaces (chapter 9); and compatibility with DOS, with Windows, and between the 16- and 32-bit versions of OS/2 (chapter 10). Chapter 11 is devoted to the general architecture of external communications and network technology rather than to OS/2-specific communications. Chapter 12 examines issues OS/2 must face in the future, including open systems models, UNIX and SAA compatibility, support for multiprocessing, and enhanced security. The material in chapters 11 and 12 is primarily tutorial, with some space devoted to OS/2-specific responses that the authors expect as a result of the issues identified or technology expected.

The index is large (29 pages) and includes all OS/2 API calls, which should be useful to developers. The bibliography is adequate and current, and the illustrations, while not profuse, are adequate. Every chapter contains a set of exercises that elicit short answers or essay responses rather than quantitative results.

The authors have written a clear, relatively complete description of OS/2 at the implementation level. I recommend this text highly for OS/2 developers or others with a specific interest in how OS/2 works. It is not a suitable primary text for a user- or design-level college course in operating systems. It is, however, an ideal supplementary text for either course and could be used for a specific course on OS/2.

The book’s greatest strengths are the blending of theory with practical implementation for a specific operating system, the complete treatment of mutual exclusion in chapter 7, and some of the insights into the future of OS/2. Its description of communications and network features is too brief, tutorial, and generally disappointing. It does a good job of describing the many UNIX features found in OS/2. It also provides excellent conceptual and implementational treatment of threads, but stops short of describing weaknesses associated with mutual exclusion at the thread level. In summary, the authors make a highly useful contribution to the literature, which will probably be widely used by the community of OS/2 developers.