Re: [stack] R

that link is really REALLY slow.  if you're a member of ACM you can
get it there -- a better instance than the PDF of the photocopy at
the link i provided.  or you can purchase the collected turing award
papers, in which it appears.  or you can probably find that paper
and others by iverson at jsoftware.com. 

here's an illustration.

given a vector v (list, array, whatever) of integers and a maximum
value n, chop v into maximum subvectors s.  intuitively, you build s
by taking as much as you can from v without exceeding n, then take 
some more, &c.  the remainder (if any) is the final subvector of s.

  v:3 2 2 0 1 4 0 4 3 2 3 4 3 4 1 1 2 2 2 4
  n:10

  bp[n;v]
0 5 8 11 13 18  / breakpoints of v 

so:

  bp[n;v]_v     / chop v at those breakpoints
3 2 2 0 1
4 0 4
3 2 3
4 3
4 1 1 2 2
2 4

i picked this one off the k mailing list -- a recent query by a k
newbie.

as it happens, bp is idiomatic.  it crops up in lots of contexts.
for example, wordwrap:

  wordwrap:{" "sv'bp[x;count each w]_w:" "vs y}

  wordwrap[20;"split words on 20 column boundaries - produce a list of strings"]
"split words on 20 column"
"boundaries - produce a"
"list of strings"

that is,

    w:" "vs y              / split the string into words
    b:msv[x;count each w]  / find the breakpoints where count <= x
    s:b _w                 / chop the words into sublists at those points
    " "sv s                / fuse the words into a string
    
array languages evolve in the direction of finding a better range of
datatypes and a better set of primitives, with the result that a larger
set of problems come in range of array solution.  programs get smaller,
have less cruft, run faster, and take less space.

here's bp:

    bp:{-2_(1+a bin a|x+-1_0,a:sums y)\[0]}

enchilada is an array-oriented concatenative language, but very very
different from APL or j or k.  i don't know why it slipped my mind
when i raised the subject.

----- Original Message ----- 
From: "Christopher Diggins" <[email protected]>
To: <[email protected]>
Sent: Thursday, March 20, 2008 2:41 PM
Subject: Re: [stack] R


> On Thu, Mar 20, 2008 at 8:02 AM, Stevan Apter <[email protected]> wrote:
>> ----- Original Message -----
>> From: "John Nowak" <[email protected]>
>> >
>> > On Mar 19, 2008, at 8:54 PM, Stevan Apter wrote:
>> >
>> >> if that's intellectual curiosity, i am marie of roumania.
>> >
> 
> My original post (sorry if sounded like I was trolling) was because
> the R language looked to me like PostScript with several overloaded
> operators for dealing specifically with arrays. I was wondering if
> that is all that it took for a language to be considered an array
> language.
> 
>>
>> www.jdl.ac.cn/turing/pdf/p444-iverson.pdf
>>
>> still the best exposition of the central ideas of array programming.
> 
> Thanks for the link. I should emphasize that I am not in any way
> trying to negate any of the benefits of array programming, but as a
> language designer I want to know what characteristics a langauge
> should have to be considered an array language.
> 
> Thanks again,
> Christopher
>