For years the dictionary only had the tree diagrams.  The equivalences
were introduced by me after 2004.  At the Hoboken JUG in January 2007
it finally hit me that KEI probably had intentionally omitted the equivalences.

The implementation of fork is not going to change any time soon.
On that particular point, at one time the C implementation was 
something like  
 return g2(f1(a,w),h1(a,w));  
where a and w are the left and right arguments (that's an in-joke, 
alpha and omega, for you APL fans).  But that could give different 
answers depending on the C compiler.  The implementation was 
changed to read
 y=h1(a,w);  x=f1(a,w);  return g2(x,y);
and remains thus to this day.

BTW, today (December 17) is Ken's birthday.


----- Original Message -----
From: Henry Rich <[EMAIL PROTECTED]>
Date: Monday, December 17, 2007 17:53
Subject: RE: [Jprogramming] Aiming at a multi-core future
To: 'Programming forum' <[email protected]>

> > For example, at the first Hoboken JUG, Roger 
> > mentioned that he had found a "bug" in the Dictionary.  
> > Here's the problematic sentence:
> > 
> >     (f g h) y   <==>  (f y) g (h y)
> 
> Good Lord, this isn't a 'bug' to be cast aside, it's a description
> of function, and one that I have counted on for years.  A 
> change of
> this magnitude should certainly make it to the Release Notes.
> Did I miss it?
> 
> If what you say is true, I have to revisit a large section of my
> code to see what the ramifications are.  Ay-yi-yi, this is 
> a headache.
> 
> Can I at least beg for advance wsrning if the interpreter is ever
> actually changed to execute the fork in other than right-to-left 
> order?
> Henry Rich


----------------------------------------------------------------------
For information about J forums see http://www.jsoftware.com/forums.htm

Reply via email to