Path: utzoo!utgpu!jarvis.csri.toronto.edu!mailrus!cornell!uw-beaver!uw-june!pattis
From: pattis@june.cs.washington.edu (Richard Pattis)
Newsgroups: comp.edu
Subject: Re: is turing the best next language to use in intro classes?
Summary: Turing <> Pascal + Modules
Keywords: turing, Pascal, CS curriculum
Message-ID: <7287@june.cs.washington.edu>
Date: 16 Feb 89 16:29:26 GMT
References: <1059@oswego.Oswego.EDU> <1989Feb13.182342.2022@ziebmef.uucp>
Distribution: comp
Organization: U of Washington, Computer Science, Seattle
Lines: 36

In article <1989Feb13.182342.2022@ziebmef.uucp>, lea@ziebmef.uucp (Simon Peter Lea) writes:
> 
> 
>     In my opinion, to the average CS undergrad, Turing is just Pascal with
>   modules.  Very similar to Turbo Pascal (4.0 on PC and MAC) - which is of
>   course similar to Modula, etc.  Having used it for the past three years 
>   myself as an undergrad here at U of T, I can vouch for its elegance 
>   and how simple it is to learn initially....

I'd like to suggest that Turing is a much better thought out language than
either Pascal or Modula-2.  It may look like both of these with a few changes
and additions, but the similarities at the syntactic level hide the big
advantages of Turing at the semantic level.

Take parameter modes.  They look just like P/M-2, but there is a big semantic
difference.  They behave like Ada's IN, IN OUT modes.  That is all parameters
are transmitted in O(1); the restriction on non-VAR parameters is that the
body of the subprogram cannot change them (checked and flagged at compile
time).

So what's the big difference?  Look in a random Pascal book at binary
searching of arrays.  In 50% of the books you will see the array passed as
a non-VAR parameter, which means that it will be copied, which means that
the subprogram runs in O(N) not O(Log N) - if the authors implement it
recursively, its O(N Log N).  Why do authors choose non-VAR, because the
array doesn't change.  Pascal forces you to choose a mode in an awkward
way.

What happens in P/M-2 if you accidentally omit VAR.  The item is copied,
changed internally, and then goes away - causing very strange problems at
run time.  This cannot happen in Turing; forgetting VAR and trying to change
the parameter yields a compile time error.

So there is more than meets the eye in Turing.

Rich Pattis