I am having trouble parsing your sentence:

> As this appears to be not a decision, it does need a rationale either.

which immediately follows

> So IMO it has nothing to do with "in J, compound (multi-
> dimensional) indices are addressed with boxes".

Once I understand what you are saying, perhaps I can
respond (if a response is required).



----- Original Message -----
From: "R.E. Boss" <[email protected]>
Date: Tuesday, August 31, 2010 5:21
Subject: Re: [Jprogramming] Splitting an integer into its digits
To: 'Programming forum' <[email protected]>

> > Van: [email protected] [mailto:programming-
> > [email protected]] Namens Dan Bron
> > Verzonden: dinsdag 31 augustus 2010 1:42
> (...)
> > 
> > If you're wondering why _1 _1 doesn't give the bottom-right 
> corner of the
> > array, like f[0][0] (or maybe f[0;0]) would give the top-left 
> corner of a
> > 2D
> > array in a C-like language, the answer is: in J, compound
> > (multi-dimensional) indices are addressed with boxes.
> > 
> > There are many ways to present the rationale for this 
> decision, but an
> > easy
> > one is: conceptually, a compound index is a single, atomic 
> thing, yet has
> > structure (this seems contradictory until you work through 
> some examples,
> > e.g. "the bottom-right corner" of a 2D array).  In J, 
> atoms with
> > (arbitrary)
> > structure are represented with boxes.  Voila.
> > 
> (...)
> 
> 
> I disagree with your reasoning.
> 
> The only reason for the behavior of { is its rank. So the answer 
> to "why _1
> _1 { (NB. the { !(reb)) doesn't give the bottom-right corner of 
> the array"
> is: the left hand rank of the dyad { is 0  .
> Compare this to dyad {.  which has a lh-rank of 1 :
> 
>    _1 _1 {. i.4 5
> 19
> 
> So IMO it has nothing to do with "in J, compound (multi-
> dimensional) indices
> are addressed with boxes".
> As this appears to be not a decision, it does need a rationale either.
----------------------------------------------------------------------
For information about J forums see http://www.jsoftware.com/forums.htm

Reply via email to