www.btinternet.com/~adrian.larner/review/davies
Database Systems
Paul Beynon-Davies
Macmillan Press Ltd, 1996, £16.99 282pp softbound
ISBN 0-333-63667-8
A Review by Adrian Larner for the Computer Journal
A DATABASE BOOK REVIEW
These are lean days for students, so the economically priced and – shall we say? – distinctively coloured textbooks of the Macmillan Computer Science Series are a godsend; and not least this volume on Database Systems. Its material is pertinent for at least the last two taught undergraduate years: it introduces the key concepts and covers data models, traditional and high fashion; DBMS functions, including front-end tools and distributed DB; DB design, including object modelling; and new developments, comprising parallel computing (in a chapter contributed by the series editor, Frank Sumner), so-called "intelligent" databases (mostly Datalog, perhaps for those that want to found their join traps securely in logic), and complex applications such as geographical information systems.
Is all this possible in less than 300 pages? Of course not, but the author makes a bold and not entirely unsuccessful attempt, aided by his pleasantly fluent writing style, which is informal without ever being casual. Each chapter has a modest (hooray!) list of references. Exercises (with sample answers) are provided. The index is useful. But the subject is too big: high time to split it into abstract data theory, specification, and independence on the one hand, and data storage and access management on the other.
Most of the book tells, as we would expect, the conventional story, even including the hoary old tale that the word “data” is a plural (page 1, belied by the author’s usage on page 11). But at least it tells the story quite well, although showing signs of inadequate pre-publication review: “mathematical sets must be named unambiguously” (no, they cannot all be named, there not being enough names to go round); “[a] direct consequence of the entity integrity rule is that duplicate rows are forbidden” (no, it is the other way round: forbidding duplicate rows means that we can impose entity integrity, i.e. every table will have some candidate key; presumably the same inversion underlies the definition of first normal form, which appeals to the concept of primary key); “relational algebra is ... procedural ... because it demonstrates the property of closure” (no, being procedural and exhibiting closure are quite orthogonal); there are minor errors in some of the SQL queries.
The author (like the rest of us) struggles to explain the inexplicable: “there is some confusion about what it actually means to be object-oriented.” Subclass/superclass (specialisation/generalisation) relationships are quite well described (although, as it happens, unnoticed, no object instances can be of two object classes one of which is a subclass of the other, whereas entity instances can be of such classes). Aggregation relationships are analysed at some length, using Wilson, Chaffin, and Herrmann’s dubious taxonomy (ship is member of fleet, unless it is the fleet’s flagship, in which case it is not member but component; “no location can be separated from the area of which it is part”, so Slough never was in Berkshire). But we never learn what semantics are supposed to distinguish aggregation from other relationships (there are none).
How can such a student text be written today? One can hardly omit the modish object-orientation (inviting death by anonymous referee); and yet its concepts are largely opaque, and in parts it is grossly mistaken; and Dr Beynon-Davies perhaps has (he would say, as part of the aggregate that is himself) a personality too eirenic even to contemplate a forthright dismissal of this upstart fashion. He is also, by the way, a thoroughgoing Platonist, thus the unconventional plural table names (“Students”) and “the phenomenon that students take ... modules ... [is] a relationship between classes”. So if, in your naïve, nominalistic way you think that “takes” relates students to modules (i.e. instances to instances), you have a lot of grinning and bearing to do. But it is just about worth it.
Copyright © 1997, 2001 Adrian Larner. The author asserts all moral rights.
The decorative image of a key (cc004239.gif) used on this page was obtained from IMSI's MasterClips/MasterPhotos© Collection, 1895 Francisco Blvd East, San Rafael, CA 94901-5506, USA.