> Finally, an anecdote. I recently bought a Samsung Galaxy tablet. Wonderful toy, except that it does not display APL characters, at least not "right out of the box". Do y'all really want to deal with that kind of headache?
Reading that from Roger, I had a flashback to 1973. I was in IBM UKSC Peterlee, doing a joint project with ASDD Mohansic on the ill-fated monster IBM FS (Future Series) project. Mohansic insisted on developing all the prototypes in APLSV. They had Selectrics with APL golfballs, hundreds of them -- the noise was deafening. I felt it was my own head banging away on that platen. We had nice quiet 3270s (CRT display units) with EBCDIC-only (conventional chars). Back then IBM didn't use ASCII. UKSC was a can-do sort of a place, so I had to learn APL on-the-job. APLSV had separate customizable byte-mapping tables for input and output. I learned APL without the benefit of the char set. Using both upper and lowercase alpha plus spare punctuation it was possible to shuffle around a mnemonically expressive pair of I/O tables. I can't remember how, except that iota was i and epsilon was e. Of course. So I learned APL, from scratch, in my own personal latin transliteration. It felt perfectly natural because I knew no different. When I got printouts from Mohansic I transcribed the code by hand and then I could understand it. Eventually a 3270 appeared (3275?) which could display APL chars. It was a long while before I took to them and persisted in using my old I/O tables for all my creative work. A curious occupational deformity? Undeniably. But it's left me with a unique viewpoint on the APL char set question. 40 years on... plus ça change -- plus ça la même chose. On Wed, Apr 10, 2013 at 6:06 PM, Roger Hui <[email protected]>wrote: > The answer to the last question in your message is, probably not. :-) The > design process of APL/J can be illustrated by a couple of stories, and a > couple of papers. The paper are: > > - *The Evolution of APL > <http://www.jsoftware.com/papers/APLEvol.htm>, *especially > the Transcript of Presentation. > - *A Personal View of > APL<http://www.jsoftware.com/papers/APLPersonalView.htm> > * > > And the stories: > > - http://keiapl.org/anec/#sort > > One of Ken’s masterstrokes in J was to redefine the dyad grade so that > x/:y is x indexed by the grade of y , which means that to sort y you > can say y/:y . Sometime later in a presentation Chris Burke showed a > list of uses of the adverb ~ (reflexive/passive). Now the monad f~ y is > defined to be y f y , and in Chris’ list he had /:~y to sort y . Ken was > startled when he saw this. > > (He didn’t know?! I would give Ken the benefit of every doubt and assume > that he probably did know. But suppose he really didn’t know. That would > make the design of x/:y even more incredible.) > > A subsequent extension imposed a total array > ordering<http://www.jsoftware.com/jwiki/Essays/The_TAO_of_J> in > J, which means that the monadic and right domain of /: are all the > arrays, whence /:~ sorts all arrays — any rank, any type. > > - http://keiapl.org/anec/#principles > > Different people claiming to follow the same broad principles may well > arrive at radically different designs; an appreciation of the actual > role > of the principles in design can therefore be communicated only by > illustrating their application in a variety of specific instances. It > must > be remembered, of course, that in the heat of battle principles are not > applied as consciously or systematically as may appear in the telling. > — > *The Design of APL* <http://www.jsoftware.com/papers/APLDesign.htm>, > 1973 > > > Now regarding the introduction of special symbols. The J ASCII spelling > was designed specifically for the 7-bit ASCII character set, as the > *APL\360 > * notation was designed for the technology available at the time. I feel > sure that if Ken were designing a language for UNICODE, he would do it > specifically for UNICODE, and not just map the J ASCII spelling to the > nearest UNICODE equivalent. Of course, there is nothing to say that you > have to approach the problem the way Ken would have. > > Finally, an anecdote. I recently bought a Samsung Galaxy tablet. > Wonderful toy, except that it does not display APL characters, at least > not "right out of the box". Do y'all really want to deal with that kind of > headache? > > > > > > > On Wed, Apr 10, 2013 at 9:05 AM, Steven Taylor <[email protected]> wrote: > > > On 10 April 2013 15:41, Ian Clark <[email protected]> wrote: > > > > > I've spent a lot of time nibbling away at the edges of this very > problem. > > > > > > sometimes it makes sense: http://lesscss.org/. Sometimes not: > > https://github.com/kripken/emscripten/wiki. For some crazy reason this > > one > > appeals to me: http://trydecaf.org/. ... and this is just scratching > the > > surface ;-) > > > > Whenever someone get's deeply into language design, it's a good thing. > > There's lots to discover that you only realise through doing... > hacking... > > or even (perish the thought) designing. There's something about APL / J > / > > K that I don't get. Is it true that a lot of these innovations occurred > in > > short time periods? I've heard a few comments and got that sense now and > > then. If so, we're not dealing with evolution or design by committee, > but > > something else that hopefully Catherine will help us understand better > > through her documentary. Maybe someone could write a book about, "Is > there > > hope for mortals?" > > ---------------------------------------------------------------------- > > For information about J forums see http://www.jsoftware.com/forums.htm > > > ---------------------------------------------------------------------- > For information about J forums see http://www.jsoftware.com/forums.htm > ---------------------------------------------------------------------- For information about J forums see http://www.jsoftware.com/forums.htm
