I can't disagree with your contention about how people (try to) approach J or other programming languages.
For my part, when I started J, I found the DoJ impenetrable and eschewed it for the first year or two of my career. I learned J through the collective wisdom and generosity of these Forums, combined with innumerable conversations (ok: arguments) with the interpreter. Once I already knew J, the DoJ became valuable (in fact, invaluable). In other words, I got to J the same way Arthur Rubinstein directed the tourist to Carnegie Hall. And, as far as I know, there is no other way to get there. -Dan PS: I'm not trying to be glib. Like any project, our documentation can always be improved and never be "finished". And over the years many intelligent, generous people have expanded and contributed to it in many different voices, media, styles, and methods. And I'm sure that will continue. But I'm equally sure there is no silver bullet (which is the flip side of "our documentation can never be finished"). Ultimately, it comes down to the individual aspirant. -----Original Message----- From: programming-boun...@forums.jsoftware.com [mailto:programming-boun...@forums.jsoftware.com] On Behalf Of Ian Clark Sent: Thursday, November 29, 2012 3:16 PM To: programm...@jsoftware.com Subject: Re: [Jprogramming] @: and capped fork > What provoked your doubt? In learning J, I've stumbled from one false assumption to another. Mainly about verb composition. Were I to be the only person to experience this, it could be put down to a weak intellect. But if my experience is not uncommon, it points to a need for J documentation to be better. Just today, whilst testing an explication utility I have on the chocks, I thought I'd stumbled on yet another unwarranted assumption I'd been making in the code. It was one I could not easily check from the standard reference material. To be sure I needed to ask on the forum. The problem with J documentation as it's structured, is that to compose 2 verbs you naturally think of a conjunction -- and J has a wealth of those. Far more than APL. If you want to choose a suitable conjunction you can read its description in Voc. But hooks and forks, especially Capped Fork, compose verbs too, yet they're not documented in a comparable way. You have to be au-fait with the Dic chapter on Trains, or Chapter 40 of JforC. It's my contention that experienced programmers learn a new language by reading the reference manual as they go along, one feature at a time. They don't digest an armful of primers or textbooks up-front, no matter how exciting and readable. Learning J, that approach just happens to induce a blind spot over Capped Fork -- and verb-trains in general. Ian On Thu, Nov 29, 2012 at 5:44 PM, Dan Bron <j...@bron.us> wrote: > Right. For this reason (or similar ones, like when g is a train), I > phrase the identity as ([: f g) ↔ f@:(g) . > > As to where this is stated: well, it's recorded informally in > innumerable documents and J learning materials. If you're looking for > formal guarantees in canonical material (the DoJ), you'll have to > arrive at the equivalence through a chain of logic. > > First, we have the the definition of capped fork in §II.F, following > the definition of non-capped fork [1]: > > (A) "If f is a cap ([:) the capped branch simplifies the forks to > i. g h y and > ii. g x h y" > (B) "The ranks of the hook and fork are infinite." > > Then, we have the definition of @: in the vocabulary [2]: > > (C) "@: is equivalent to @ except that ranks are infinite." > > Which refers back to the definition of @, which is given in the > Vocabulary as [3]: > > (D) " u@v y ↔ u v y" > (E) "x u@v y ↔ u x v y . > > So, after adjusting for the different names given to the verbs, we the > following correspondences: > > (Ai) "([: f g) y ↔ g h y" vs (D) "f@g y ↔ f g y" > (Aii) "x ([: f g) y ↔ g x h y" vs (E) "x f@g y ↔ u f g y" > > (B) "The ranks of the hook and fork are infinite" vs > (C) "@: is equivalent to @ except that ranks are infinite." > > Which, as far as I can tell, establishes the identity ([: f g) ↔ f@:g > (provided we heed the advice given in latter's definition, i.e. > "because a conjunction applies to the entity immediately to its right, > expressions to the right of conjunctions commonly require > parenthesization.") > > What provoked your doubt? > > -Dan > > [1] §II.F, definition of trains > http://www.jsoftware.com/help/dictionary/dictf.htm > > [2] Vocabulary entry for @: > http://www.jsoftware.com/help/dictionary/d622.htm > > [3] Vocabulary entry for @ > http://www.jsoftware.com/help/dictionary/d620.htm > > -----Original Message----- > From: programming-boun...@forums.jsoftware.com [mailto:programming- > boun...@forums.jsoftware.com] On Behalf Of bob therriault > Sent: Thursday, November 29, 2012 12:00 PM > To: programm...@jsoftware.com > Subject: Re: [Jprogramming] @: and capped fork > > HI Ian, > > If your v includes an adverb such as / the long left reach of > conjunctions could get you into trouble. That would be part of the > parsing rules for verbs vs conjunctions. > > (+:@:+/) 3 4 5 > 42 > ([:+:+/) 3 4 5 > 24 > > Cheers, bob > > On 2012-11-29, at 8:49 AM, Ian Clark wrote: > >> Department of Sudden Doubts... >> >> If u and v are verbs, do (u@:v) and ([: u v) really behave the same >> under all circumstances? >> >> If so, where would I go to find this fact written up? >> --------------------------------------------------------------------- >> - 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 ---------------------------------------------------------------------- For information about J forums see http://www.jsoftware.com/forums.htm ---------------------------------------------------------------------- For information about J forums see http://www.jsoftware.com/forums.htm
