I think Devon's analysis is probably correct (given that I was the coder :-)).
I came to J from an APL background where there was no Rank. The primitive
<each> was used quite a lot instead. J defines the idiom &.> as "each" and &>
as "every" and I naturally gravitated to them rather than Rank initially. I
imagine I still have that bias to some extent.
Incidentally, {.&> doesn't save anymore brackets than {."0
> From: Devon McCormick
>
> You are correct - the effect is the same as '{."0' which you can see if
> you substitute this: the expression returns the same result. It's
> probably a matter of how the expression was arrived at or how the coder
> is used to working, though I would maintain that the rank-zero take is
> a clearer expression of what is intended.
>
> On Wed, May 13, 2009 at 10:10 PM, gary ng <[email protected]> wrote:
>
> > On Wed, May 13, 2009 at 6:53 PM, bill lam <[email protected]>
> wrote:
> >
> > > On Wed, 13 May 2009, gary ng wrote:
> > > > Are there any hidden 'gem' or reason to say '{.&>' rather than
> > > > '({."0)' as to me the only reason for the &> is to modify the
> > > > rank of {. ?
> > >
> > > If the argument is un-boxed, why would you use {.&> ? What effect
> > > did you expect from that > ?
> > >
> >
> > That is what puzzled me. This is the original code I came across :
> >
> > (,|....@}:)(,.|....@}:"1)({.&>~#{.i:@#)(65}.a.)(>:@i.{.[)'E'
> >
> > from http://www.craigmurphy.com/blog/?p=1417#comment-66863
> >
> > It does save a pair of '()' but that seems to be not a good practice
> > as it doesn't use the 'primary' functionality of '>' which is
> > to unbox
----------------------------------------------------------------------
For information about J forums see http://www.jsoftware.com/forums.htm
