A Quarter Century of Unix is a history of Unix, a kind of annotated collection of reminiscences. It begins at the “birth” of Unix, with Ken Thompson looking for a machine to play Space Travel on, then jumps back to provide the context, both in the history of computing in general and in the particular setup at Bell Labs. Part two describes the work done up to 1974, both on Unix and on the tools and language (C) so closely associated with it. Part three tries to pin down some of the things that made Unix unique: its style, the strong contributions by users and user groups, and the key role of some of its more famous tools. Parts four and five trace the expansion of Unix: the development of BSD and the commercial Unixes, the creation of SUN, the ambivalent relationship with DEC, legal issues and attempts at standardization. The final section offers an overview of the current status of Unix in its many different versions and offers some ideas about where it is heading. There is also a very brief glance at some of the systems that it has influenced, including Bell Lab's new Plan 9 system. The finale has Dennis Ritchie, Brian Kernighan and others offering their ideas on what made Unix work. Particularly noteworthy is the solid treatment of legal issues (three chapters altogether) and the coverage of events outside the United States (in Australia, Europe and Japan).

The format of A Quarter Century of Unix, with most of the text in the form of extended quotations, runs the risk of discontinuity and lack of focus. Salus has chosen and edited his source material well, however, and inserted his own summary and exposition in appropriate places. The result is a great read, with the voices of the various creators providing unique perspectives on the events they participated in (some scores of people are quoted from at length).

I did spot a few minor inconsistencies in the text—on page 155 we read “It was 32V that became 3BSD in 1979”, though the Unix versions tree on page 61 shows no such influence—and errors—on page 253 we have “It was clear that AT&T hadn't objected to other derivatives: Linux, MINIX, etc. In the autumn of 1988...”, implying that Linux existed in 1988 (and Linus' name is misspelled in the index, too). But these are just quibbles. A more weighty criticism would be that the book sometimes reads more like myth than history, with the participants portrayed like epic heroes. (It's rather obvious that Salus himself is a Unix fan.) This may worry the historians, but in a way it is the legends and myths that are the most influential, so the distinction is perhaps moot.

A Quarter Century of Unix doesn't assume specialized knowledge, but the more you know about Unix (and to a lesser extent, about architectures and operating systems) the more you will get out of it—if you've never used awk, for example, you will probably have little interest in reading about its origins and development. The main audience will be programmers, administrators, and users with extensive Unix experience. Historians and sociologists of the computer industry will find Salus' work an essential source of primary material, and marketing types might well learn a thing or to from it. A Quarter Century of Unix should be a great success; it's just unfortunate that it wasn't written years ago!