HYPERMEDIA AND THE WEB:An Engineering approach. David Lowe and Wendy Hall. Wiley. 623pp. Pounds .50. - ISBN 0 471 98312 8.

Thirty years ago there was a series of software disasters. As a result software engineering was born: the aim has been to evolve a disciplined approach to software construction, so that software products are delivered on time, meet their budget, are easy to maintain and are pleasant to use.

Today there is still a series of software disasters, usually funded by the taxpayer and involving some supposedly prestigious contractor. It is arguable whether these failures occur because basic software engineering principles are not applied, or because software engineering is itself inadequate. It is obvious, either way, that software engineering still has some way to go.

It is wrong, however, to be too negative. Alongside the laughable disasters there are at least as many successes, many of them software products we use daily - often without knowing it.

³ÉÈËVRÊÓƵ

The production of hypermedia and of web presentations is at the same stage as software production was 30 years ago. Products are often thrown together, rather than built using a disciplined process, and a lot of them do not work very well. In particular, people are now finding that maintaining such products is becoming hopelessly expensive.

This book aims to do in the hypermedia area what software engineering has done for software as a whole. Clearly it is extremely timely. Equally clearly it is only the start of what will be a long and never completely successful struggle. The problem the book tackles is well set out in John B. Smith's excellent foreword: "Hypermedia is changing from a cottage industry of hand-crafted pages to a spectrum of activities that culminates in enterprise-wide systems. The intuitive, trial-and-error methods that have worked in the past do not meet today's needs, and much less those of tomorrow. As the authors state, what is needed is an engineering approach."

³ÉÈËVRÊÓƵ

Smith also captures the essence of the book's approach: "A key issue of hypermedia engineering is to develop an abstract representation of the application that can be scrutinised, tested and ultimately implemented."

The book tackles its task by first discussing fundamentals in depth (over some 200 pages), and then presents desirable practices (some 130 pages). These practices relate to software engineering areas such as analysis, design and verification, and to hypermedia areas such as information structuring and link design. Finally there is a discussion of research developments, again taking some 200 pages.

Clearly, in a new discipline, a lot of the material is going to be leading-edge research, but I would like to have seen more material relating to current practice; it is disappointing, for example, that the case study is short and hypothetical.

The book is nicely presented, with a useful set of goals presented at the beginning of each chapter. It is a reasonably easy read but, perhaps inevitably, is rarely exciting. (The most successful text on software engineering also wins prizes from students as their most boring course book.) It does not focus exclusively on the worldwide web, but nearly everything it says relates to web materials.

³ÉÈËVRÊÓƵ

Often the authors' approach is to take existing software engineering practice and see how it applies in the more specialised area of hypermedia. Surprisingly, there is little focus on object-oriented methods, although these are a centrepiece of many software engineering books. There is a good set of references, though few are dated later than 1996, apart from work associated with the authors.

Overall the book is a good and comprehensive start towards creating a new and important sub-discipline. In 30 years' time we shall undoubtedly have hypermedia creations that are unimaginably bigger than today's. This book will help make that happen.

Peter Brown is professor of computer science, University of Kent at Canterbury.