Manuscripts Wanted Ed Shaw Renaissance Data Systems P a r k West Finance Station P. O. Box 20023 New York, New York USA 10025-1510 Telephone: 212-864-3078 Renaissance Data Systems seeks manuscripts of booklets (20-50 pages) suitable for use in the classr o o m wherein A P L is used as the expository notation or computer implementation. This request is a result of the workshop discussion at A P L 9 2 chaired by Professor H o w a r d Peelle. A number of participants decried the dearth of A P L materials available for use in the classroom. To satisfy this need, R D S hereby solicit ¢ material or an expression of interest and commitment to developing such material. Royalties will be paid to author ¢ should the publications become commercially successful. Material should include a careful development of the subject matter and a collection of problem sets suitable for independent study. If you are interested in this opportunity, please contact Renaissance Data Systems. If you would like a catalog of all books and soRware in A P L and J currently available from RDS, please send a self-addressed, stamped envelope to the above address. ¢ I a m aware of its advantages. But, can all operations be reduced to parallel algorithms? Or, can all programs be implemented reasonably without the use of GOT0s? A case in point, try writing a totally parallel program that figure ¢ the averages and standard deviations for N r a n d o m set ¢ of data, all of which have a random number of elements. As for pipelining processors such as the i486 and i586, I personally don't k n o w h o w such processors process interpreted languages. But, the compiled code that constitutes the A P L interpreter and environment has m a n y GOT0 statements put into it at the machine-language level by the compiler. As a result, adding GOT0s to your A P L code just slows your program down. O n the other hand writing ungodly code and doing handstands just to avoid GOTOs is just as bad, maybe even worse--there are more statements to parse, interpret and execute, not to mention it will take a long time to write! In this case I think w e should be judicious with our use of GOT0. A good idea would be to follow the ideas given in Jim Lucas' article, "Making A P L Cheap," and try to eliminate possible "beer bellies" caused by the injudicious use of GOTO. Anoher w a y to m a k e A P L cheap and speed code at the same time is to develop more efficient compilers on micro or mainframe computer ¢. With the marked increase in supercomputing, I don't see w h y someone cannot write a compiler that will compile several dialect ¢ of A P L for a family of pro.cessors using a 3090 or even a Convex. The resulting machine language would then be downloaded to the target machine. Sincerely, - Todd G r u h n Department of Chemistry Michigan State University East Lansing, Michigan 48825 U S A Letter to the Editor To the Editor: I enjoyed Mr. Van Batenburg's article "L-OneTwo-Three (/;1 :...L2 :...L3 :) Considered Harmful" [van Batenberg, F.H.D.; A P L Quote Quad, 21 (4), August 1991, 330-337] very much. I t h i n k he may have a point regarding the use of control structures in APL. However, two questions I ask, and I hope t h e y stir debate in the APL community are: 1. Do we really need GOTO (÷)? 2. How much does unharnessed GOTO affect a function's execution? Having w r i t t e n my share of spaghetti code I know the difficulties of debugging and maintaining such code Also, having w r i t t e n totally parallel code Correction R o b e r t G. B r o w n Executive Editor, APL Quote Quad 116 Bentley Avenue Old Bridge, N J 08857-~1336 USA Due to an oversight during preparation, the M a r c h 1992 issue of A P L Quote Quad did not credit Joseph D e K e r f with work done on the 2991 Index to A P L Ar~cles. Mr. DoKerf was a co-editor of the Index. We regret this omission and acknowledge his considerable contribution to this project. ¢ APL Quote Quad

/lp/association-for-computing-machinery/correction-fo5DLvr6H0