¢o*,~.w~,1 • Ckmc~m'ts V ~ 4 pp 351-352 PerllSm~Pre~l.~L 1971 P n m e d m G r ~ t l l n m L
structure On the other hand, according to Backus, the fact that the 704 had only three index registers was a coincidence, and not a factor when the designers imposed a hmlt of three subscnpts on arrays APL, m contrast, began m the late 1950's as a means of commumcatJon w~th people, not at first with _=. ~.~. .-.-.- . machines Iverson used it m his classes at Harvard when descn'bmg topics such as linear programming, sortmg, and the structure and operation of machines Although already pubhshed m book form m 1962 (K E Iverson, "A Programming Language". John Wiley & Sons), APL attracted a new audience when used to descnbe the detailed behavior of the newly announced System/360 m Apnl 1964 (A D Falkoff, K E Iverson. and E H Sussenguth, 1964, "A Formal Descnptlon of ° System/360". IBM Systems Journal v 3, p 198-263) ~ -. ,. , ° Because APL uses a rich set of symbols not available on card punches, It probably seemed difficult at first to Implement Indeed, amongst programming languages it has the distraction of being used at the desk and on the blackboard independently of the existence of any computer Implementatmon. At just the nght time, however, IBM developed the 1050 terminal and the No one who was fortunate enough to attend it is likely to Selectnc typeball It was therefore possible to provide a forget the Los Angeles Conference on the History of font and to communicate mteracttvely m APL w~th a Programming Languages, held m June under the auspices computer This was accomphshed on the IBM 7090 m of the Associatmon for Computing Machinery (ACM) 1965 Falkoff and Iverson have discussed the influence of After Grace Murray Hopper's fascinating keynote ad- the terminal keyboard and type-ball on the development dress on early history, the program took two and a half of the language, and what msespecially interesting is then days to pass through the spectrum from FORTRAN to recogmtlon that unwelcome physical constraints APL. The languages represented at the Conference were frequently resulted tn stgmficant improvements and all created and m use by 1967, they remain m use 10 generahzattons m the language Early computers had memories so small that great skdl years later, and each has had considerable mtluence on the field of computing The speakers (each of whom was and ingenuity was required to use them at all For one of the desmgners) attempted to tell why they created example, the first computer in Pomona College's Geology the languages tn the ways they did, what purposes they Department had only 32 words of memory, the programs intended, and what constraints were unposed by machine being on wired boards Yet, minute though this is by today's standards, we accomphshed useful work with It tmplementatmon The hlstonc nature of the conference became evident But so much attention had to be given to the efficient use at the very first session, when John Backus introduced of early machines that we became myopic, concentrating Cuthbert Hurd In 1953 Hurd had approved Backus' on hardware and losing sight of the evolution of matheproposal to develop a practical automatic programming matical notation that preceded the computer age and stdl system for the IBM 704 That. of course, was the birth of continues FORTRAN The last session of the Conference was At the Los Angeles Conference, John Backus, who remarkable for the discussion by Fred Brooks, a pnn- was manager of the ongmal Programmmg Research clpal architect of IBM System/360, of the paper by A D Group responsible for creating the FORTRAN language, Falkoff and K. E Iverson on "The Evolution of APL" said that all conventional languages, such as FORTRAN The influence of machine structure on language desmgn and ALGOL, "create enormous, unnecessary intellectual should interest a geologist, because his busmess is with roadblocks m thinking about programs These 'Von evolution and rots controlhng factors For example, Neumann' languages constantly keep our noses pressed because the 36-brat word of a 704 can encode six charac- m the dmrt of address computaUon and the separate ters, each occupying six bas, FORTRAN restricted the computation of single words, whereas we should be length of names to six characters, and continued to do so focussing on the form and content of the overall result even when mmplemented on later machines of dttferent we are trying to produce the fact that such languages
COMPUTER CORNER by Donald B. Mclntyre
CAGEOVol 4 No 4-C
351
352
Computer Corner
have dommated our thmlung for twenty years is symbols, which included the St Andrews cross, x, munfortunate because their long-standmg famdlanty troduced for muluphcatJon m 1631 wall make tt hard for us to understand and adopt new Developments m computer technology, such as the programming styles which one day wdl offer far ~eater avadabflity of megabyte memories and virtual storage on intellectual and computational power" what we used to call mini-computers only a year or two Although Backus did not menUon APL m his talk, ,t ago, both reqmre and make posssthle the style of proanswers his descnption of a programming style nsmg . grammmg that is charactensUc of APL Its symbolism above *'the separate computation of smgle words" and helps us grasp the intellectual content of an algorithm without extraneous and irrelevant matters prescn'bed by possessing far greater intellectual power than do the convenuonal languages. It can be truly stud that its a machine Bertrand RusseU stud "A good notauon has a symbols **make reasoning easter", R ts a language to subtlety and suggesuveness which at tunes make at seem think m If one objects, as some do, that the notation ,s almost like a live teacher" Slowly, like acceptance of unfam.har and too complex for nonspeclahsts, we Ougtred's symbol for muitJphcahon, people in many fields are recogmzmg that APL provides the good notashould remember the words of Wdham Oughtred, tion we need pioneer m mathemaUcal symbolism Wnung m 1647, he .said " Which Treat,se being not wntten m the usuall synthet,cal manner, nor with verbous expressions, but m the inventive way of Anahtice. and with symboles or notes of things instead of words, seemed unto many very SUGGESTEDREADING hard, though indeed it was but their owne d:~dence, being Backus, J, 1978, The history of FORTRAN I, 11. and Ill. m Preprmts ACM SIGPLAN History of ProgTammmg scared by the newness o[ the delivery, and not any Languages Conference, Los ,Angeles, California, June I-3, dl~iculty in the thing it sel[e For th,s spec,ous and 1978 ACM SIGPLANNouces, v 13, no 8. p 16%180 symbohcall manner, ne,ther racketh the memory w,th Backus, J. 1978, Can propammmg be hberated from the Von mult,phc,ty of words, nor chargeth the phantaste w,th Neumannstyle '~ A functionalstyleand its algebraof programs Comm ACM, v 21, no 8, p 613-641 companng and laying things together, but phunly Falkoff, A D and lverson,K E, 1978,The evolutzonof APL, m presenteth to the eye the whole course and processe of Prepnn's ACM SIGPLAN Hzstory of Programming every operation and argumentation" And also in 1632 Languages Conference. Los Angeles, Calffornm, June I-3, "This manner of setting downe Theorcmes. whether they 1978 ACM SIGPLANNoUces, v 13, no 8, p 47-57 be ProporUons, or Equations, by Symboles or notes of Goldsteme, H H, 1972. The computer from Pascal to Von Neumann Pnnceton Umv Press. 378p words, is most excellent, arttfic,all, and doctrmall lverson. K E., 1972. APL m exposlUon IBM Tech Rep 320Wherefore I earnestly exhort every one, that desweth 3010 (avmlablefrom APL Press, Box 378, Pleasantvflle.New though but to Iooke into these noble Sc,ences MathemaYork, 10570) hcall, to accustome themselves unto ,t. and mdeede it ,s lverson, K IL, 1973, An introducUonto APL for sctenusts and engineers IBM Tech Rept 320-3019 (available from APL easie, being most agreeable to reason, yea even to sence Press. Box 378. Pleasantvdle.New York. 10570) And out of this working may many singular consectarles Mclntyre. D B, lntroductmnto the study of data matrices m be drawne, wh,ch wtthont this would, it may be, for Fenner. P. od, Modelsof geologic processes--an introduction ever lye hid . ". In these passages Oughtred was ' to mathematicalgeology Am Geol Inst. Washington.D C, P encouraging h,s readers not to be afrmd of his new DB'M-A-I-44, B-I-23, C-I--4