Newsgroups: comp.lang.scheme
Path: cantaloupe.srv.cs.cmu.edu!bb3.andrew.cmu.edu!newsfeed.pitt.edu!gatech!swrinde!howland.reston.ans.net!ix.netcom.com!netcom.com!NewsWatcher!user
From: hbaker@netcom.com (Henry Baker)
Subject: Re: expressiveness and strong type system?
Message-ID: <hbaker-1403961017520001@10.0.2.15>
Sender: hbaker@netcom20.netcom.com
Organization: nil organization
References: <OZ.96Mar14014447@nexus.yorku.ca>
Date: Thu, 14 Mar 1996 18:17:52 GMT
Lines: 32

In article <OZ.96Mar14014447@nexus.yorku.ca>, oz@nexus.yorku.ca (ozan s.
yigit) wrote:

> somewhere in the current discussion jeff dalton repeats an occasionally
> heard claim that stronger type systems result in "less expressive" languages.
> this is an interesting claim, but it occasionally gets stuck like a fuzzball
> in the throat of further inquiry. perhaps people who have thought about this
> issue can provide some examples? references to any formal work? is there any
> support for felleisen's neat "conciseness conjecture" for example?

I presume that by 'stronger type systems', you mean statically typable systems.

The problem with most type systems is that they insist on going it alone, with
no help from the programmer.  Since this choice results in type systems that
have recursive algorithms, it necessarily restricts the expressiveness of
the language.

One can conceive of type systems which can be given help by the programmer,
and therefore allow a much greater degree of expressiveness, while allowing
the type check to run in some reasonable amount of time.

Many of the proposals for 'run-time' typing are 'really' a mechanism for
the programmer to help the (static) type system.

A concrete example --
I contend that a type system that cannot type implementations of the Y
combinator restrict the expressiveness of the language.

-- 
www/ftp directory:
ftp://ftp.netcom.com/pub/hb/hbaker/home.html

