Path: utzoo!attcan!uunet!seismo!dimacs.rutgers.edu!rutgers!cs.utexas.edu!samsung!dali.cs.montana.edu!uakari.primate.wisc.edu!zaphod.mps.ohio-state.edu!sdd.hp.com!hplabs!otter.hpl.hp.com!otter!phmb
From: phmb@otter.hpl.hp.com (Peter Brooks)
Newsgroups: comp.theory
Subject: Re: Hashing help needed
Message-ID: <3140004@otter.hpl.hp.com>
Date: 18 Jul 90 07:32:32 GMT
References: <2385@uop.uop.edu>
Organization: Hewlett-Packard Laboratories, Bristol, UK.
Lines: 23

>The Godel numbering scheme is fine, but the message says to use p1 for
>computing the modulus, and p1, p2, p3 for the Godel number. It should be clear
>that the two p1s are different. That is, use some function such as p1^L * p2^W

Yes, my mistake, it should have been p1,p2,p3,p4. Sorry for any confusion -
I know which p I meant :-). The method you propose is indeed easier for
the computer - the reason I particularly like the Godel method is the knowledge
that before you mod anything you do have a unique number so you can be
sure of where your synonyms are coming from. This is, of course, a quite
spurious argument from the point of view of elegance (I do agree that the
actual arithmetic is not that elegant, but it never is). I do sometimes feel
that aesthetics are too often left out in the rush for simplicity. I also
have belief that taking a general approach in the initial stages of solving
a problem leaves you with a vastly more flexible result.

A small justification for the above belief is the frequency with which
large software projects with a cast of thousands produce monsters which
(fortunately) are usually still born or die young. Whilst pieces of hack
code cobbled together to solve a very specific 'one off' problem are
modified and embellished over the years and evolve until their origins are
well obscured - vide UNIX (tm).

Peter Brooks