"Ha, I’d forgotten about that. I remember having fun with it at the time."
I gave a clue: "courtesy of Dan" when I mentioned the exercise :) "It’s also different from Ken’s original suggestion to use ]. (or, in today’s terms, 2 : ‘v’)" Incidentally, in wicked Jx, dex (].) is defined as dex=. ]conj where conj is the counterpart of adv (conj is _2?:0 and adv is _1?:0 ). Within conj ] replaces ]. and operates directly on the verbs ( just as verbs within adv operate directly on the argument verb); moreover, one can continue to shift to higher levels within higher levels and sometimes this can be somewhat confusing just as expression with many nested parentheses can but they are just as expressive). In other words, verbs (adverbs and conjunctions) are first-class citizens. That is all for now because it is time for the NY J Meeting. Yes! On Tue, Apr 12, 2016 at 10:49 AM, Dan Bron <[email protected]> wrote: > Pepe wrote: > > [0] [Jprogramming] Tacit adverb definitions? > > http://www.jsoftware.com/pipermail/programming/2006-July/002627.html > <http://www.jsoftware.com/pipermail/programming/2006-July/002627.html> > Ha, I’d forgotten about that. I remember having fun with it at the time. > > Not having to parenthesize the LHA to a conjunction under certain > conditions (the pertinent condition being insight under consideration in > this thread) came as a lovely little revelation when I was just hacking > around in the session manager. > > I can’t remember what I was playing with at the time, but I do remember > that little observation being an enjoyable outcome. It might have been when > I was exploring someone’s assertion (Randy’s, maybe, because of his deep > experience with APL?) that J evaluates strictly left-to-right. > > Anyway, it was in the era when I started to look at J’s grammar directly > (i.e. trying to actually understand $II.E) after spending years developing > a subconscious intuition for it. > > I’m still not nearly as conversant with J’s grammar, or parsers [in the > compiler sense] in general as Raul is, nor do I think I expect I ever will > be. I “get” it (as in, J’s evaluation of sentences I type rarely takes me > by surprise), but I’d have to put in some effort and thought in order to > explain it to a third party. And I would probably be forced to explain it > in more human / intuitive terms than what is spelled out in the parse table. > > In the same vein, though I don’t think it was at the same time, I also > remember a sense of satisfaction when I learned that 3 : ‘label_.’ (a > label-less label*) could be used as a minimal and semantically-transparent > J equivalent to APL’s diamond or C-like languages’ semicolons within > explicit definitions. > > That is, label_. acts as a logical line separator that forces the > expression on the left to be evaluated to completion before the expression > on its right is executed. > > This is distinct from normal labels, because if you wanted to use those, > you’d have to vary the label names if you wanted more than one line > separator in a single definition, which works, but feels like a hack. > > It’s also different from Ken’s original suggestion to use ]. (or, in > today’s terms, 2 : ‘v’), because while that works for (the majority of) > nouns and verbs, it does “the wrong thing” when one side or the other is an > operator (i.e. an adverb or conjunction). That is, its scope is limited. As > punctuation, label_. works for any wordclass. > > -Dan > > * https://en.wikipedia.org/wiki/The_Gateless_Gate < > https://en.wikipedia.org/wiki/The_Gateless_Gate> > > > ---------------------------------------------------------------------- > For information about J forums see http://www.jsoftware.com/forums.htm > ---------------------------------------------------------------------- For information about J forums see http://www.jsoftware.com/forums.htm
