One thing I do not like about explicit definitions is unexpected side effects:
m=. +/ .* inc=: 1 :'&.(%&m)' 3 inc NB. OK &.(%&3) + inc NB. WHOA! &.(%&(+/ .*)) inc=. (%&) (<'&.'`) (`:6) 3 inc &.(%&3) + inc &.(%&+) Somebody said recently that J has no reserved words, of course, with the exception of the most useful: m, n, u, v, x and y (if you are programming explicitly). I understand and share your frustration about the (`) related bug: inc ((%&)(&.`))(`:6) However, while I was preparing my presentation for the conference I discovered that using a pro-conjunction for (`) alleviates the problem: l=. ` inc=. (%&) (<'&.'l) (`:6) inc ((%&)(<'&.'l))(`:6) On Fri, Aug 24, 2012 at 5:53 PM, Raul Miller <rauldmil...@gmail.com> wrote: > Oh, hmm.. > > But if I use > > to =: [ + 1 i.@+ -~ > > it seems to work: > > 5 10 to 15 20 > 5 6 7 8 9 10 11 12 13 14 15 > 10 11 12 13 14 15 16 17 18 19 20 > > or maybe not: > > 5 10 to 12 20 > 5 6 7 8 9 10 11 12 5 5 5 > 10 11 12 13 14 15 16 17 18 19 20 > > I probably want fill elements there? If so, I should probably use to > with an explicit rank. > > Anyways, for now anyways, I think that that |length error is a good thing. > > As for the |domain error, I think I'd prefer to fix that in inc: > > inc=: 1 :'(&<.)(&.(%&m))' > > Finally, my problem with ((<'&.'`)(`:6)) is not that I do not know > what it is doing. I know exactly what it is doing, in the sense that > I can sit down and walk through the creation of intermediate results, > building up the gerund that will be interpreted as a train. And, I > know why (<'&.'`) is equivalent to ((<'&.')`) and I know why this > construct, when it is used, swaps the verb derived from (%&) off to > the right of &. when the gerund being built is interpreted as a train. > > My problem is that I do not have good mental labels to use when I > think about it those elements. I think that I'm lacking abstractions, > for reasoning about that kind of expression. I do not have good > words, either, for talking about expressions built into gerunds which > are auto-interpreted as trains. And, I suppose, a part of my problem > is that J has problems displaying some of these intermediate results. > I dislike working in a way that's likely to trigger interpreter bugs. > > -- > Raul > > On Fri, Aug 24, 2012 at 5:28 PM, Jose Mario Quintana > <jose.mario.quint...@gmail.com> wrote: >> There are a couple of problems with >> >> to =: [ (+ i.) 1 + -~ >> >> 5 10 to 15 20 >> |length error: to >> | 5 10 to 15 20 >> |[-0] >> >> >> 5 to (0.27 inc) 8 >> |domain error: to >> | 5 to(0.27 inc)8 >> |[-0] >> >> >> inc=: 1 :'&.(%&m)' >> >> Is the explicit equivalent (to some extent) to >> >> inc=. (%&)(<'&.'`)(`:6) >> >> I wrote the second form because the original requested a "tacit form" >> admittedly referring to a verb but I like to be "tacitus" and in some >> respect I also would like to be as Tacitus "... and as well as the >> brevity and compactness of his Latin prose, he is known for his >> penetrating insights ..." >> >> The "unnamed parentheticals" are adverbs forming trains in a typical >> tacit adverbial programming fashion. Dan Bron wrote a pretty good >> general explanation: >> http://www.jsoftware.com/pipermail/programming/2010-November/021172.html >> >> >> On Fri, Aug 24, 2012 at 11:46 AM, Raul Miller <rauldmil...@gmail.com> wrote: >>> On Fri, Aug 24, 2012 at 11:17 AM, Jose Mario Quintana >>> <jose.mario.quint...@gmail.com> wrote: >>>> to=. [ + i.@:>:@:<.@-~ >>>> >>>> 5 to 8 >>>> 5 6 7 8 >>> ... >>>> inc=. (%&) ((<'&.'`)(`:6)) >>>> >>>> 5 to (0.25 inc) 8 >>>> 5 5.25 5.5 5.75 6 6.25 6.5 6.75 7 7.25 7.5 7.75 8 >>> >>> I like your argument patterns. >>> >>> But I am uncomfortable with your definitions (especially the unnamed >>> parentheticals) -- perhaps because I do not understand them well >>> enough to reason about them generally? >>> >>> Anyways, here's how I might define these words: >>> >>> to =: [ (+ i.) 1 + -~ >>> inc=: 1 :'&.(%&m)' >>> >>> Thanks, >>> >>> -- >>> Raul >>> ---------------------------------------------------------------------- >>> 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
