BOOK REVIEWS

My most serious objection is that there is no distinction for contentsand function for System/360 in contrast to System/370.

Several tables enumerate the instructions set for System/370.Simtlar tables may be found in the Principles of Operation. Only atoken description is provided. Can you imagine the complex com-mand BXLE being allocated a single sentence?

3) Systems and Devices: I found this part most useful. Here thespecific features of each computer model are described. There isenough detail provided to give the principles of the new features. Iwould like to see detail at the register level, but this is strictly my ownprejudice.

There is a particularly good discussion of buffer storage and its con-trol. It makes clear how buffer storage expedites the acquisition ofdata and instructions that have been recently referenced, or whereimmedite neighbors have been recently referenced.

There is only a token description of I/O devices. My curiosityabout the 3330 disk system was not satisfied.

IVAN FLORESBaruch CollegeCity Univ. New YorkNew York, N. Y. 10010

B72-27 The Fourth Generation-International Computer State ofthe Art Report/#I. (Maiden Head, Berks., England: Infotech,Limited, 501 pp., $95.00, subscriptions available for series of reportsto be issued.)

The book is composed of three major parts: a dialogue on thefourth generation theme, a series of papers (called presentations), andfinally six invited papers.

Of the three parts the "presentations" part is by far the best.The dialogue presented in the first part is constructed by the

editors by excerpting the papers. This editorial device has recentlybecome very fashionable; however, it is very difficutlt for the editorsto excerpt avoiding distortions and nonsequiturs. The result is thatthis "cute" technique does not lend itself to expressing serious andinsightful thought.

The "Presentations" part is composed of the following twelvepapers.

"The Influence of Transaction Processing on Fourth GenerationSystems," by G. E. Hall. Attempts poorly to convey the requirementof a specific user.

"The Fourth Generation from the User's Viewpoint, " by I. M.Laing and J. R. S. Kistruck. States some interesting requirements fordata base addressing.

"Fourth Generation Computer Design," by G. G. Scarrott. Givesa uniquely insightful picture of the evolution of computing systemsand the key role software and hardware interfaces play in keeping thesystem complexity to a human scale. Interfaces such as the channellevel interface allow organizationally separate groups to develop aportion of the system such as the main frame I/O controller and aperipheral processor in a relatively independent mode.

"Addressing Structures: The Focal Point of Fourth GenerationArchitecture," by N. D. Gammage. Focuses correctly on one of thecentral architectural issues of the fourth generation and does so with avery lucic discussion of the notions underlying paging, segmentation,and codeword systems.

"The Nature & Implications of Software Engineering," by J. N.Buxton. A verbose and not particularly insightful contrast of com-puter sciences versus software engineering.

"Machine-Independent Operating Systems: A Functional Ap-proach to Design," by P. Cox. A number of somewhat trivial pointsare made (e.g., OS should be clean and tight) with excessive use ofjargon.

"Time Sharing Management Software: Man & Machine," byD. F. Hartley. Discusses in an interesting manner a very neglectedand important area, i.e., the management of the installation as a com-plement to system design.

"The Nature & Benefits of Modular Operating Systems," byD. Morris. Gives the reader a good understanding of the underlyingfundamental issues of a process oriented OS while presenting a designunderway at the University of Manchester. In spite of the fact thatsome aspects of this specific design are questionable, reading thepaper is a very valuable experience.

"Benefits for User and Producer of an Engineering Approach toCompiler Design," by D. Hendry. Pleads for engineering disciplineand standards in software development.

"Network Systems: The Economic Solution for the Fourth Gen-eration User," by I. M. Barron. A set pf arguments (mostly economicand marketing) why IBM will not release a fourth generation line ofsystems and why the rest of the industry would not either.

"Technical, Market & Investment Factors Promoting & Re-straining Change in Machine Architecture, " by J. G. Laski. Not veryinformative.

"The Interaction of Software Engineering & Machine Structure,"by C. Strachey. Use of high-level languages is mandatory for develop-ment efficiency. Thus we must learn empirically to optimize the ma-chine architecture for support of the chosen high-level language.

The third and final part of the book contains the following invitedpapers.

"Concern for Correctness as a Guiding Principle for ProgramComposition," by E. W. Dijkstra. Basic theme is. we have to learnhow to put structure in programs so that their correctness can beestablished logically.

"A Survey of Store Management Techniques," by C. A. R. Hoareand R. M. McKeag. Surveys a number of page replacement algo-rithms with no mention of the role of LRU and OPT (min) algo-rithms.

"Systems Measurement: Theory & Practice," by K. Kolence. Acommercial for Boole & Babbage.

"Macros and Their Uses," by P. J. Brown. A tutorial on macrosand their use. It addresses a very neglected area. Advocates a generalpurpose macro preprocessor for a fourth generation system.

"On Reliable & Extendable Operating Systems, " by B. W. Lamp-son. A fairly detailed design description of a proposed operating sys-tem.

"Manageable Software Engineering," by R. W. Bemer. Severalinteresting notions concerning the management of software develop-ment. Stresses the importance of suitable development environment(quick turn-around facility).

U. 0. GAGLIARDICen. Res. Comput. Technol.Harvard Univ.Cambridge, Mass.

1459