🚀 go-pugleaf

In article <23950@megaron.cs.arizona.edu> gudeman@cs.arizona.edu (David Gudeman) writes:

   But there is another approach to universality.  The existence of
   English plus current programming languages plus all the specialized
   notations of mathematics, logic, linguistics, chemistry, and many
   other areas show that a fairly universal language is possible.  I
   don't think there is anything inherently wrong with saying that all
   this is one language.  Of course it is not a very cohesive or elegant
   language.

I've been for quite a while working with a language that allows
"plug-in" modules for various purposes. The most interesting of the
modules are full-blown applications, and I normally think of this as
an "applications integration language". However, some of the modules
are extensions to give the language notations from realms other than
those it was implemented for.

   Given this view of what a universal language is, the question becomes
   ``to what extent can this hodgepodge be unified and streamlined?''.
   The question becomes a quantitative rather than a qualitative one.

Here's where either I lose what you're talking about, or you lose
touch with what I perceive the problem to be.

The problem isn't streamlining the various specialized notations
together, it's providing all the facilities for manipulating those
objects that are going to be needed. For example, even though I have
the appropriate notations to work with the field in question, if I
need to manipulate lists of objects and the underlying language
doesn't provide primitives for list manipulation, then it may be
easier to re-implement the notations in a language that supports lists
than vice-versa.

   This unification is not a process of throwing out the specialized
   features of various languages, rather it involves finding
   relationships among the various specialized features, subsuming them
   with more general features.

Yes, but if that "more general" feature doesn't map simply to the
specific features needed, it's not clear you've made an improvement
over including them both. Of course, the LISP solution is to include
all three cases. But I think that that way lies madness, at least on
the scale of a universal language.

   Unix has shown pretty conclusively the benefit of combining different
   languages. [...]
   Even so, I don't think anyone would call this collection a single
   language.  But maybe with enough ingenuity something similar can be
   designed that is so cohesive that it _could_ be called a single
   language.

It's called PERL :-). But there are still domains that it doesn't work
well in.

   Maybe a universal language
   would involve a lot of little sublanguages, but all fully integrated
   into a single entity, so that you are not committed to a single
   sublanguage just because you chose to start in that language.

Such a language is doomed to failure. The problem can be expressed as:
Do you have hooks for dealing with sigournism, a field of study whose
subject matter will be discovered in 2012, and spring into existence
thereafter? Likewise, the xplee data structure (patent applied for in
1996) is a critical facility - how do you intend to incorporate it?
And after you've added them, you've just created a "more universal"
language than your previous "universal language."

English solves this problem by being "infinitely extensible"; it's
allowable to create new terminology that combines smoothly (or
roughly, if that's the way you want it) with the rest of the language,
and can be used for discussing sigournism or xplees. Of course, there
are a number of "infinitely extensible" programming languages.  None
of them seem to be taking over the world, and Herman isn't interested
in them when they're pointed out to him as solutions to his problem.

	<mike
--
Love and affection,					Mike Meyer
Of the corporate kind.					mwm@relay.pa.dec.com
It's just belly to belly,				decwrl!mwm
Never eye to eye.