If I recall correctly the equivalence was, x (c a) y <-> (x c y) a
where (x c y) a meant, at least for the interpreter, that the product of (x c y) was passed, as the argument, to a. Furthermore, the interpreter (mistakenly) allowed to pass adverbs and conjunctions to the adverb a (again, if I recall correctly). On Wed, Mar 16, 2016 at 11:21 AM, Pascal Jasmin <[email protected]> wrote: > Could someone remind me what > > > (a c) and (c a) used to do as trains? (in J5) > > was (c c) ever allowed? > > (sorry this is difficult to search) > > > > > ----- Original Message ----- > From: "[email protected]" <[email protected]> > To: [email protected] > Sent: Wednesday, March 16, 2016 10:12 AM > Subject: Re: [Jprogramming] Unbox request for requests > > [For historical issues, it helps to spell out "Version" and "Release" > with the numbers: > > J Version 1 was rather known as "APL\?" or "APL90 J" as > presented in http://www.jsoftware.com/papers/J1990a.htm#intro > > J Version 2 - 7: = 1990-1993 > > J Version 7 (open source) was then followed by the polished, > commercial, non-source "J Release 2". The never was something > called "J Release 1". (You are free to view the preceeding > versions as the "Release 1s", but it is in no way an official > designation.) > > J Release 2 ...: = 1994 and later > ] > > Henry Rich asked: > > How far back do you have to go to find an old system that allows an > > adverb as an operand to an adverb? > > The AA bident is still working, and the question should rather be: > > "How far do you have to go back to find an old system > which does *not* allow the AA bi-dent?" > > > 7 didn't allow it, did it? > > J Version 7 allowed it: > > neitzel 422 > j7 > J7 Copyright (c) 1990-1993, Iverson Software Inc. All Rights Reserved. > > 0!:0 'uname -a' > ULTRIX ips 4.4 0 RISC > > APLscan =. /\ > + APLscan 1 2 3 4 5 > 1 3 6 10 15 > > as did > > neitzel 426 > j > J5.1 Copyright (c) 1990-1992, Iverson Software Inc. All Rights > Reserved. > > which is the oldest version I have still running and which supports AA. > > > The AA bident was first introduced with J Version 3.3: > > Changes in Version 3.3, 1991 06 25 > [...] > a-trains implemented > c-trains implemented > > (I cannot find the AA bident yet in the blue "Tangl.Math + DoJ" booklet > which must be Version 2.9ish or even 3. It is listed in the J Version 4 > DoJ.) > > While I saved the CS department's decsystem from being scrapped and still > operate it 24/7, my even earlier J Versions were binary-only (DOS), and > almost vanished along with the my office's PC-XT which mainly served > as terminal hooking into our Unix systems. > > But I found a "floppies.tgz" from that time with J Version 2.9, > fired up "dosemu", and so here is a "pre-AA J:" > > https://www.gaertner.de/~neitzel/img/no-aa-j29.jpg > > > The list of possible trains was the same for J Version 5.1 - J Version > 7. J Release 2 added further ones, such as A0 V1 V2. > > J Release 5.1 [2002] did away with the more complex bi/tridents: > > * All tridents (dictionary Section II F) other than the > verb-verb-verb case no longer work. > * All bidents (dictionary Section II F) other than the verb-verb, > adverb-adverb, and conjunction with an argument cases no longer > work. > > IIRC, the reason stated for the removals was "are in the way for improved > debugging". Did it improve? > > (I am asking because I didn't bother to do anything with J Release 5.x/6.x > myself for licensing and src reasons. I remained a happy camper with > J Rel. 4, and still run it on more obscure systems such as the Sharp > Zaurus.) > > Martin Neitzel > > ---------------------------------------------------------------------- > 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
