Date: Sun, 22 Jan 1995 20:23:00 EST From: "Rob Slade, Social Convener to the Net" Subject: "Object Databases: the Essentials" by Loomis BKOBJDTB.RVW 941213 "Object Databases: The Essentials", Loomis, 1995, 0-201-56341-X, U$27.95 %A Mary E. S. Loomis %C P.O. Box 520, 26 Prince Andrew Place, Don Mills, Ontario M3C 2T8 %D 1995 %G 0-201-56341-X %I Addison-Wesley Publishing Company %O U$27.95 416-447-5101 fax: 416-443-0948 %P 230 %T "Object Databases: The Essentials" Object technology, and the associated object-oriented programming, is anecdotally reported to be a major tool in the evolution of software and information technology development. The object model is said to provide for faster development, greater code testability and reliability, and greater code reuse. The association with graphical user interfaces and "point and click" environments adds a perception of "ease of use" to the mix. Object technology, however, requires a very different perspective from that which has driven the information systems of the past. This is particularly true with regard to database management. Loomis states that the book assumes only "minimal" previous knowledge about either object technology or database management. I very strongly suspect, though, that the reader's perception of the book will be radically different depending upon whether the reader has a background in objects, databases, or neither. Those with an object background will see the book as a vindication of the superiority of the object model. They will fail to note that the work speaks of the importance of distributed programming and access to existing data. They will likely also fail to note that the book buries admissions that object models so far fail to address these issues. Those with a traditional DBMS background will find a very unconvincing argument for the superiority of objects. Loomis states that object identifiers are not keys and that traversal paths are not pointers, but fails to prove her statements. The definition of "object databases" as those with strong links to C++ and Smalltalk is not likely to clarify the topic for this group. Those who have little of either background will likely be bemused. The "object model" is not laid out in any depth until chapter six, and the basic concepts are never presented in a manner that would be lucid for newcomers. The claims for object databases are interesting, and it is likely that they are more flexible than the relational structure. However, they also conflict with some very basic concepts, such as data independence. The relational model is subject to many years of study, and mathematical models for correctness. No such background exists for objects. Those from the traditional DBMS world should study this work although it will likely give only a vague idea of either the possibilities or pitfalls of objects. A fuller understanding will likely require hands-on experience in attempting to make object databases do what relational databases won't. copyright Robert M. Slade, 1994 BKOBJDTB.RVW 941213 ============= Vancouver ROBERTS@decus.ca | Lotteries are a tax Institute for Robert_Slade@sfu.ca | on the arithmetically Research into rslade@cue.bc.ca | impaired. User p1@CyberStore.ca | Security Canada V7K 2G6 |