Browse Books

Reviewer: Ralph R. Martin

The aim of this book is to provide a bridge between traditional graphics standards such as GKS and PHIGS and the approach of object-oriented systems such as Smalltalk that also include graphics capabilities. The author points out both the advantages of the former, such as portability and device-independence, and the additional flexibility and extensibility of the latter. His intent is to convince graphics standards makers to incorporate object-oriented ideas in the next generation of graphics standards. After a clear introduction that sets out the authors main ideas, the book contains a brief summary of Smalltalk80, a discussion of interactive user interface architecture in object-oriented systems, and a description of graphics in Smalltalk. Next, the author considers GKS and object-oriented methods, graphics model hierarchies and PHIGS, and GEO++, his own object-oriented multilevel graphics system. Finally, he discusses examples in PHIGS and GEO++, the benefits of inheritance for GEO++, other approaches such as prototypes and delegation, the implementation of GEO++ in Smalltalk, constraints and related ideas, and further considerations for an object-oriented standard. The early part of the book is often pedestrian and contains basic material that has been presented elsewhere. The book is restricted to two-dimensional graphics and concentrates on the graphics packages without enough emphasis on how they will be used in an overall application architecture. The author also emphasizes his own work with GEO++. The book could profitably include much more discussion of and comparison with other object-oriented graphics standards and systems such as NeWS, MacAPP, and HyperTalk. Several technical difficulties beset this book, including poor typesetting in a difficult-to-read typeface, several missing or erroneous references, and rather too many niggling errors in the English. Also, sadly for a book on graphics, many of the earlier diagrams are not particularly helpful or informative. Despite these criticisms, the author has something useful and important to say. Exaggerated claims have sometimes been made for object-oriented approaches, but Wisskirchen talks good sense and has thought clearly about the issues involved. The examples provided by his own system clearly show some of the benefits to be obtained, and I hope this work will have some influence on the standards makers.