Path: utzoo!utgpu!news-server.csri.toronto.edu!mailrus!wuarchive!zaphod.mps.ohio-state.edu!samsung!munnari.oz.au!goanna!ok
From: ok@goanna.cs.rmit.oz.au (Richard A. O'Keefe)
Newsgroups: comp.lang.misc
Subject: Re: The Universal Language (Was Re: Efficient Fortran)
Message-ID: <3543@goanna.cs.rmit.oz.au>
Date: 10 Aug 90 08:36:23 GMT
References: <24013@megaron.cs.arizona.edu>
Organization: Comp Sci, RMIT, Melbourne, Australia
Lines: 31

I seem to be missing something in all this discussion.
I thought we _had_ computation-universal languages already.
To make an glaringly obvious point, if machine code isn't
"universal enough", what are you going to translate your
universal language _into_?

A group in the States did some really serious work on natural language
processing some years ago, in Fortran, on a PDP-11.  (As I recall it,
they had lots of little Fortran programs exchanging data.)  I've heard
of a project (controlling a CAT scanner, if I recall correctly) that
write a large object-oriented program, in Fortran.

To give you a natural-language parallel:

	in the tongue that we speak, all thoughts we can form can be
	said in words that have but one sound in them.  One may have
	to use far more words than one would like, but each time that
	I have tried to say what I meant in words that have but one
	sound in them I have found that I could do it.  It may be
	hard to think of good words for such a form of speech, but if
	you know this tongue well, you can do it.  Is this not true
(*)-->	of the scripts we use to make tools that count do what we want
	as well?  What do the folk who ask for a way to write all
	scripts want, but that it be less hard than now?  Why must a
	way to write scripts that is less hard to use be a way that
	is so big that it is hard to learn?  Is not Scheme a good
	way to write scripts for tools that count?	

(The starred line means "computer programs".)
-- 
Taphonomy begins at death.