Newsgroups: comp.lang.misc
Path: utzoo!utgpu!tmsoft!mason
From: mason@tmsoft.uucp (Dave Mason)
Subject: Re: Can Turing survive?
Message-ID: <1989Jan12.031706.26492@tmsoft.uucp>
Followup-To: comp.lang.misc
Reply-To: mason@tmsoft.UUCP (Dave Mason)
Organization: TM Software Associates / Ryerson Polytechnical Institute, Toronto
References: <117400001@uxa.cso.uiuc.edu> <2658@ficc.uu.net>
Date: Thu, 12 Jan 89 03:17:06 GMT

From my reading of the Turing, Numerical Turing, and Turing Plus
documentation, and Mini-Tunis Operating System, and writing a few
Numerical Turing programs, my answer is "I certainly hope so!"
Numerical Turing in particular really impressed me (a compiler & O/S
hack) as extremely nice for developing Numerical Analysis type programs.

Turing Plus is a considerable improvement on my (and a fair number of
other people's) previous favourite concurrent programming language:
Concurrent Euclid (developed by the same folks).

This semester I will be teaching my Operating Systems course using the
Mini-Tunis O/S and Turing Plus.  The student group project will be
porting Mini-Tunis to run on a bare virtual VM/370 machine.  If anyone
has questions about how the students relate to & like the language,
send me mail in a month or so.

In article <2658@ficc.uu.net> peter@ficc.uu.net (Peter da Silva) writes:
>I just read the turing article. I don't know... the syntax is full of little
>arbitrary glitches (for example, using a token for the last character of a
>string that can't be stored anywhere),
Just like C?

> the seperate compilation handling
>is pretty hairy (if submodule A accesses something in a program that isn't
>passed to it, that program has to be available when the module is compiled),
No, a stub file of the program has to be available.  How else can you
have a type safe compiler (which Turing claims to be (Turing-Plus has
escapes where required))?

>and the I/O handling is built in to the language (the language is otherwise
>so nice, and perfectly capable of providing the capabilities in a standard
>library... just as 'C', Modula, and so on do).
You could of course do it that way if you want, just ignore the
builtin I/O & call your own routines.  For small programs there is no
question but that it is convenient to have it built in.

>I got the impression of a more advanced teaching language, rather than a
>replacement for Modula, 'C', and so on... It's got a lot of nice ideas,
>but doesn't hang together all that well.
Whether it succeeds in displacing much of the C market is for history
to decide, but it definitely SHOULD displace Modula, it is a MUCH
nicer & cleaner language.  (It almost makes B&D look good :-).  It is
also a dynamite teaching language.  It is the only procedural/
imperative language that I think should even be considered for intro
teaching (I have a personal soft spot for Scheme, but that's a
different religious war).  And the nice thing is that it scales quite
smoothly into large scale software engineering projects!  UofToronto
teachs Turing to first year Engineers as well as Computer Science
students.  They claim to get engineers who see how computers may even
be useful to them, and even understand a little about programming!
Dramatically better results than previously achieved with Fortran to
the Engineers.

	../Dave
(Disclaimer: I am a part time graduate student at UofT, but I've NEVER
been known to let details like that get in the way of my calling
things the way I see them.  I really think Turing is a good language.)