> Backus had a reciprocal relationship with APL. 
> He admired it but didn't think it had gone far 
> enough.  In his FP he described an "insert" 
> function which was later adopted by APL, as a 
> better way to describe what was then called 
> "reduction" in APL.

I don't think this description of how "insert"
got into APL is accurate.  APL had slash and 
slash-bar many years prior to Backus' August 1978
paper.  Iverson in his "Operators and Function" 
(1978-04-26) described a innerproduct/ example, and
innerproduct slashbar was interpreted exactly the
same as how innerproduct/ is interpreted in J today.
It _is_ better to call / "insert" rather than 
"reduction" because for an arbitrary function f,
f/ may not reduce anything.  In older APLs f/ was
restricted to scalar functions, and in such cases 
f/ does reduce an axis.

Backus discussed APL in section 8 of his 1978 paper.
He erroneously described APL as having "exactly
three functional forms, called inner product,
outer product, and reduction".  As of 1978, there
were two others: bracket axis (APL\360 circa 1966)
and scan (APLSV circa 1972).



----- Original Message -----
From: Eugene McDonnell <[EMAIL PROTECTED]>
Date: Monday, March 19, 2007 8:13 pm
Subject: Re: [Jgeneral] John Backus Died

> 
> On Mar 19, 2007, at 7:53 PM, June Kim wrote:
> 
> > http://www.nytimes.com/2007/03/19/obituaries/20cnd-backus.html? 
> > _r=1&oref=slogin
> > -----------------------------------------------------------------
> -----
> > For information about J forums see 
> http://www.jsoftware.com/forums.htm
> Backus had a reciprocal relationship with APL. He admired it but   
> didn't think it had gone far enough. In his FP he described an  
> "insert" function which was later adopted by APL, as a better way 
> to  
> describe what was then called "reduction" in APL.
> 
> It was some of his cohorts who,without meaning it, benefitted APL; 
> 
> they were in charge of a computer, but chose not to give all 8 
> bits  
> of a byte. Since this was required by the APL group, they had to 
> go  
> their own way, successfully.
> 
> My first manager at IBM was Andy Kinslow, who was another one of 
> the  
> group that included Backus. With Andy, I helped create the first 
> IBM  
> time-sharing system.
> 
> Eugene
----------------------------------------------------------------------
For information about J forums see http://www.jsoftware.com/forums.htm

Reply via email to