> I still don't understand why monad { produces boxes; perhaps to be
> consistent with its (necessarily) boxed input? In this case consistency is
> a hindrance; in almost every use I've seen (including this one) we open the
> result of { in the end.
Ken designed the monad { and the dyad { were together, and
if i j k are integer arrays then (<i;j;k){y is equivalent to ({i;j;k){y .
e.g.
(<1 2;3;4 5){y=: 100+i.3 4 6
146 147
170 171
({1 2;3;4 5){y
146 147
170 171
----- Original Message -----
From: Dan Bron <[EMAIL PROTECTED]>
Date: Monday, January 7, 2008 11:59
Subject: Re: [Jprogramming] Fitting factors of positive integers
To: [email protected]
> Raul wrote:
>
> fac2a=: */@ >@, @ { @ (<@(1,*/\)/.~) @ q:
>
> Since the inverse of q: is */ you may
> reduce */@[EMAIL PROTECTED]: to f&.q: like
> so:
>
> fac2b=: >@, @ { @ (<@(1,*/\)/.~) &. (q:"1)
>
> (perhaps the inverse of q: should be changed
> to */"1 ?)
>
> I still don't understand why monad { produces boxes;
> perhaps to be
> consistent with its (necessarily) boxed input? In this
> case consistency is
> a hindrance; in almost every use I've seen (including this one)
> we open the
> result of { in the end.
>
> Moreover, the boxes conflict (in my estimation) with the spirit
> of J, where
> homogenous rectangular arrays are the norm, and boxes only used
> for the
> exceptions, where homogeneity and rectangularity is not an
> option. But, by
> definition, >@{ is homogenous and rectangular.
>
> (Of course I understand that we are now constrained by backwards
> compatibility)
----------------------------------------------------------------------
For information about J forums see http://www.jsoftware.com/forums.htm
