On Sat, 06 Jun 2009, PackRat wrote:
> Thanks, Raul, Bill, and Fraser! However, after reading your responses,
> I'm more in the dark now than before, concerning dyadic left and right!
>
> Rearranging Bill Lam's response a bit:
> > ... Firstly [ and ] are verb, in that context DOJ said they will
> > yield their noun argument. ... To be verbose, verb itself can only
> > see its immediate left or right noun arguments. ...
> > J should be easier to understand if you can avoid pre-mature shortcut
> > and work in its first principle.
>
> I am really trying to avoid what you term a "pre-mature shortcut" by
> taking the DOJ's EXACT AND LITERAL wording in its normal English sense
> without any special meanings or understandings associated with it which
> aren't stated. As I stated in my original message:
> > > Now here is where I need further enlightenment: ...
> > > Taking the Dictionary definition exactly and literally, it would
> > > appear that ... everything to the left of the "]" should be thrown
> > > out and only the "]" and its rightward elements would be returned.
>
> And you wrote:
> > Suppose a b c are verbs and you define a train f=: a [ b ] c
>
> As I said above, using the EXACT AND LITERAL wording of the DOJ meaning
> of dyadic [ and ] in the standard English sense and applying it to what
> I see in your train above, moving from right to left, the DOJ meaning
> would be to view it as ( a[b ) ] (c) , where I've parenthesized what's
> to the left and right of ] . Taking the DOJ definition at its exact
> and literal meaning, that would mean the result is c , with a[b
> being tossed out (ignored) in the process. In theory, if I gave the
Hi Harvey,
This is not DOJ said. There is no such thing as a `tacit form
calculus' that will do symbolic calculation on a tacit form alone.
So let's forget tacit form for the moment, I'm sure you know what [ or
] will do when applying to nouns, and that was the _entirety_ of what
DOJ said. There is nothing particular about [ or ] as regarding to
tacit form.
you said DOJ implied a [ b ] c will be evalulated as ( a[b ) ] (c)
This is the pre-mature shortcut. Let's defer this evaluation and
regard a [ b ] c un-evaluated. You may disagree because it contradicts
to tacit forms programming where you write a series of sentences in
tacit forms to solve problems. Bill must be wrong to say tacit forms
are not evaluated by J. Sure J will, but not by tacit form calculus.
Say if you enter a [ b ] c in session window and press enter, The
result as that will be echo in the following line is the same
a [ b ] c
So the sentence is not yet `evaluated', so will be any verb train.
(Depending on your global setting it may display as 'a [ (b ]c)' but
ignore that for the time being)
So what is a[b]c ? Is it a or c? Neither. a[b]c is a[b]c itself
because you did not provide a noun argument to [ or ] so that these
verbs [ and ] (also a b c) will not take any action.
OK you now give argument x/y to it, and evaluation starts
x (a [ b ] c) y NB. 5 verbs in a train, therefore 2 fork
x (a [ (b ] c)) y NB. v1 v2 v3 v4 v5 <-> v1 v2 (v3 v4 v5)
(x a y) [ x (b ] c)) y NB. fork
(x a y) [ (x b y) ] (x c y) NB. fork
(x a y) [ foo ] bar NB. evaluate b and c
(x a y) [ bar NB. ] not evaluated until here
(x a y) NB. [ not evaluated until here
(actual sequence of evaluating fork is undefined/parallel, but ignore
it again for the time being otherwise you'll be more puzzled)
If you cannot understand a particular tacit form, then provide it some
x or y and do evaluation step by step. After more practice, it become
a second nature and you will be able to see what a tacit form will do
without actual evaluation. At that point you will be smarter than J
parser because you can perform tacit form calculus that J parser
cannot.
Are you more in the dark now than before? ;-)
--
regards,
====================================================
GPG key 1024D/4434BAB3 2008-08-24
gpg --keyserver subkeys.pgp.net --recv-keys 4434BAB3
唐詩179 王維 奉和聖製從蓬萊向興慶閣道中留春雨中春望之作應制
渭水自縈秦塞曲 黃山舊遶漢宮斜 鑾輿迥出千門柳 閣道迴看上苑花
雲裡帝城雙鳳闕 雨中春樹萬人家 為乘陽氣行時令 不是宸遊玩物華
----------------------------------------------------------------------
For information about J forums see http://www.jsoftware.com/forums.htm
