On Thu, 14 Mar 2002, Ivan Tomek wrote:

> Hi,
> 
> Can you give an example where this ordering would be useful?
> 
> Ivan 
> 
> 

Lexicographic ordering of words (non-ASCII) in a dictionary. 
The words have to be mapped to collections. 

And Richard O'Keefe mentioned a data mining application....

HJH


 
> > -----Original Message-----
> > From: Richard A. O'Keefe [mailto:ok at cs.otago.ac.nz]
> > Sent: Wednesday, March 13, 2002 8:03 PM
> > To: squeak-dev at lists.squeakfoundation.org
> > Subject: Re: A suggestion
> > 
> > 
> > "Lex Spoon" <lex at cc.gatech.edu> wrote:
> > 	> (There are other orders which are also compatible 
> > with =, such as
> > 	>  - first compare the lengths,
> > 	>  - then compare the elements ONLY if the lengths are the same.
> > 	>  This is almost always faster, and has some nice properties.
> > 	
> > 	Gee, yet another way to compare collections.  Note, 
> > though, that this is
> > 	not a total ordering, either -- arrays of different 
> > lengths will be ~=
> > 	but they will be neither #< nor #> .
> > 	
> > No, it's a total order.
> > 	#(4) < #(1 2)		(length is less)
> > 	#(1 2) < #(4 2)		(length =, first element <)
> > 	#(1 2) < #(1 4)		(length =, last element <)
> > 
> > If two arrays have different lengths, then one of them is shorter,
> > and that one is less.  If they have the same length, then either
> > all elements are equal, so the arrays count as equal,
> > or one position has unequal elements, and the first such position
> > determines the order.  (Of course, taking the last position would
> > also work.)
> > 
> > This order was used on strings in PL/CV,
> > and it's _almost_ the order Prolog uses.
> > (Prolog actually compares function symbols first, then arity,
> >  and then elements.  So if you use f, f(X1), f(X1,X2), .. and so
> >  on for sequences, you get this order.)
> > 
> > I'm not advocating that ordering; lexicographic has always suited me.
> > I brought it up because that's the reason why I didn't just post an
> > [ENH].
> > 
> 
>