Path: utzoo!utgpu!jarvis.csri.toronto.edu!cs.utexas.edu!asuvax!ncar!husc6!husc7!carlton
From: carlton@husc7.HARVARD.EDU (david carlton)
Newsgroups: news.groups
Subject: Call for Discussion: comp.lang.functional
Message-ID: <1921@husc6.harvard.edu>
Date: 23 Feb 90 05:03:59 GMT
Sender: news@husc6.harvard.edu
Reply-To: carlton@husc4.harvard.edu (david carlton)
Organization: Harvard University Science Center   Cambridge, MA
Lines: 36


Tried to post an official call for discussion earlier... Pnews couldn't
deal with moderated newsgroups, so I sent it to announce-newgroups@wherever,
as the group creation regs told me to do.  But nothing seems to have shown
up, and it's been a while, so am just posting this to news.groups, and am
hoping that people don't mind this slight lapse of protocol.  Apologies if
the other message shows up as well...

The question has come up of what we mean by functional languages.  When
pressed for a definition of them, I come up with something like

     i) functions are first class objects.  absolutely necessary.
    ii) there really _shouldn't_ be any side effects... yes, lisp and
        scheme are in some (very important) sense functional, but once you
        start using setq/set!, they lose a certain je ne sais quoi...
   iii) and lazy evaluation is really nice, too.. though not absolutely
        necessary.
    iv) in sum, functional languages are really the lambda calculus in
        drag.

I would be willing to define functional languages for the purpose of the
group as languages which satisfy i) and ii); so languages in which
functions are first class objects, and in which there are no side effects.
This would exclude lisp and scheme (which have their own newsgroups
anyways) and ml (sorry...) but would at least allow the functional core of
the above, as well as all of your other favorite languages (hope, miranda,
haskell, etc.)  What do people think about that definition?  It seems to be
what we are converging on, anyways.

	And what about the name?  Somebody else suggested comp.functional,
on the analogy of comp.object; I myself prefer comp.lang.functional, but
what do others think?

David Carlton
carlton@husc4.harvard.edu
husc6!husc4!carlton