Newsgroups: comp.lang.scheme
Path: cantaloupe.srv.cs.cmu.edu!das-news2.harvard.edu!news2.near.net!howland.reston.ans.net!ix.netcom.com!netcom.com!NewsWatcher!user
From: hbaker@netcom.com (Henry Baker)
Subject: Re: Unix Weenies (formerly: Removing READ)
Message-ID: <hbaker-2802950920590001@192.0.2.1>
Sender: hbaker@netcom5.netcom.com
Organization: nil
References: <SCHWARTZ.95Feb27183623@galapagos.cse.psu.edu> <28Feb1995.095828.Alan@LCS.MIT.EDU>
Date: Tue, 28 Feb 1995 17:19:29 GMT
Lines: 56

In article <28Feb1995.095828.Alan@LCS.MIT.EDU>, Alan@lcs.mit.EDU (Alan
Bawden) wrote:

>    From: Scott Schwartz <schwartz@galapagos.cse.psu.edu>
>    Alan@lcs.mit.EDU (Alan Bawden) writes:
>    | OR PERHAPS WE COULD COME UP WITH SOMETHING EVEN
>    | BETTER THAN THAT!  We could make -progress- instead of continuing to use
>    | 1970's technology.
>    ...
>    What decade does scheme belong to, by the way?
> 
> I'd also put Scheme firmly in the 1970's.  This year marks the 20th
> anniversary of the birth of Scheme -- and it hasn't changed much since
> then.  Not unlike Unix, many of the good ideas in Scheme were actually
> worked out in the 1960's and before.

I don't know if that was their intent, but my understanding of the Scheme
community is that they wanted to clean up existing Lisp semantics.  Such
a cleanup job doesn't necessarily introduce new features, but refines the
ones that are already there.  One look at the Common Lisp book (god forbid
that you accidently drop the book on your foot) indicates that such cleanup is
long overdue.

Thus, Scheme (and the ANSI Lisp standardization effort) has essentially been a
backward looking project.  Nothing against that, but there has been very little
forward looking by Lisp people, except for the Lispm window stuff, the
CLOS object stuff, the early attempts by symbolics and lmi to do
OODB's, and the reflective stuff.  (Yes, I _did_ see the 'Life of Brian' :-)

Some recent advances have occured outside the Lisp world:

 * The Hindley-Milner type system has proved to be quite powerful and
quite efficient (despite its worst-case complexity).  There's nothing
here that keeps it from being useful in Lisp languages; Mitch Wand developed
a system about 10 years ago to utilize such a system in Lisp, but it
has apparently never caught on.  It isn't necessary to force everyone
to use it, but it would sometimes be nice to have it for some uses and
some programs -- for providing peace of mind (I don't care about/buy the
argument about its being more efficient at run-time).

 * The whole distributed object world seems to have left Lisp behind.

 * Although Lisp is arguably the best scripting language -- e.g., AutoCAD,
Emacs, etc. -- its use has not caught on, perhaps due to the absolutely
dreadful S-expression tools in the non-Emacs world.

 * Lisp is an excellent place for a 'visual language' to start.  In fact,
Lisp Machine Lisp should have been called 'Visual Lisp', in retrospect.
Why is READ so controversial, when it isn't even particularly important
any more (i.e., with GUI's and persistent OODB's).

------

Why someone must give up the power, simplicity and elegance of S-expressions
in order to do 'advanced research' on programming languages is beyond me,
and a bit of an oxymoron, to boot.
