Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Path: utzoo!mnetor!seismo!ut-sally!utah-cs!shebs
From: shebs@utah-cs.UUCP (Stanley Shebs)
Newsgroups: comp.lang.misc
Subject: Re: Software Reuse  --  do we really know what it is ?
Message-ID: <4673@utah-cs.UUCP>
Date: Fri, 3-Jul-87 14:34:03 EDT
Article-I.D.: utah-cs.4673
Posted: Fri Jul  3 14:34:03 1987
Date-Received: Sat, 4-Jul-87 13:53:52 EDT
References: <8706160502.AA26398@ucbvax.Berkeley.EDU>
Reply-To: shebs@cs.utah.edu.UUCP (Stanley Shebs)
Distribution: world
Organization: PASS Research Group
Lines: 30
Keywords: declarative functional logic programming modularity

Utah has a lot of functional programming fans, but they assiduously avoid
*writing* functional programs and instead hack C or Pascal or Lisp or Prolog
implementations of functional programs.  From what I can see, things are
pretty much the same elsewhere.  Functional programming will not take off
until somebody writes real programs in a purely functional style.
Once this happens, optimization will follow naturally as users clamor for
better performance.  So here is "Shebs' Challenge":

I will pay $250 to the first person who recodes the 15,000 lines of C code
in the rogue-like game "hack" into a purely functional style (modulo I/O).
Any publicly available functional language may be used, but the result must
be behaviorally identical to the C version, down to screen management,
scoreboards and savefiles.  Speed will not be considered as part of behavior,
nor will space consumption. To be eligible, the resulting program must be
made generally available, either by net posting or by distribution - it need
not be put into the public domain, however.  This offer will stand for five
years;  the two hundred fifty bucks has not been set aside specifically,
but I will be more than happy to produce the sum at any time.

Potential recoders should note that this program would be at least an order
of magnitude larger than any other functional program that has been written
(so far as I know), so the experience gained would be quite valuable and
possibly publishable.  It would also provide researchers with a sample
program that is considerably more realistic than "fibonacci" or "sieve".
This offer *is* serious; comments on ambiguities and loopholes are welcome.
When these have been ironed out, I will post the challenge in Sigplan Notices,
since there is no way to guarantee authenticity of a net message.

				stan shebs
				shebs@cs.utah.edu, seismo!utah-, youase