Lines: 58
Newsgroups: comp.constraints
Message-ID: <4d01d3$aha@luxor.mch.sni.de>
From: ifprolog@pgtr0158.mch.sni.de (IFCO)
Path: cantaloupe.srv.cs.cmu.edu!rochester!udel!news.mathworks.com!fu-berlin.de!news.belwue.de!news.uni-stuttgart.de!news.ruhr-uni-bochum.de!news.rhrz.uni-bonn.de!news.rwth-aachen.de!genesis.westend.com!news2.gtn.com!news.gtn.com!gtnduss1.du.gtn.com!ius.gun.de!roka.net!news.space.net!news.ecrc.de!news.mch.sni.de!luxor.mch.sni.de!pgtr0158!ifprolog
Subject: Re: Comparative study of CLP languages ?
Date: Wed, 10 Jan 1996 08:37:39 +0000
References: <4ctbts$b43@luxor.mch.sni.de> <DKx4Iz.J5t@ecrc.de>
X-Gateway: ZCONNECT XX ius.gun.de [UNIX/Connect v0.73]
MIME-Version: 1.0
Content-Type: text/plain; charset=ISO-8859-1
Content-Transfer-Encoding: 8bit

Micha Meier (micha@ecrc.de) wrote:

Dear Micha,

: thank you very much for yet another description of your sexy Prolog :-)
Our pleasure! 
: As far as its CLP capabilities and comparisons with other systems
: are concerned, I would like to have some more substance rather than just
: enumerating its features and saying that on some tests it is x-times faster
: than some other system. I understand that you have to sell it for your living,
: but you must also understand that this is a scientific newsgroup.
: All of us are certainly interested in fair comparisons of different systems, so you
: may perhaps invest some time and provide one which goes more in depth
: and can be repeated by others. As a matter of fact, there is much more in
: developing CLP applications than just speed of the application itself.

Of course you are absolutely right, we to would like to see more comparative
studies in this area. Unfortunately, as a commercial company we do not have
very much time available for such comparisons.

You are right to say that the main "selling point" that makes CLP "cost
effective" is coding time, where I would advocate high-level programming
as derived from Prolog ... and the power of constraints to reduce search
and narrow the problem ...

If you or some other institution would like to make a comparison of CLP tools
on UNIX / Windows then we would be happy to provide a IF/Prolog set of figures
or a system for the test period. From what I have seen so far, I think we
would fair quite well.

I do not have the time to develop "benchmarks" or "standard problems" which
represent "standard" applications - this is still new technology. But, I am
willing to contribute to getting this technology known and in use!

Comparing programming languages though is somewhat harder. People on the whole
know C and Prolog is new - they then compare their most recent C application
with their first Prolog application. Despite this , Prolog is being used increasingly
in industry and I can provide some excellent references where Prolog has
shone through.

In closing, don't forget industry in this newsgroup, constraints have shown
that they have serious commercial implications and have been used by several
firms very successfully. A standard set of "benchmark" problems, rather than
programs, would also help us to develop better commercial systems and show
the performance in comparison.

It might also help this research field develop in a useful practical way rather 
than delving off into some fashionable area, in a few years.

Discussion welcome.

Andrew

Dr. Andrew Verden

tel. : +49 89 7936 0037                            IF Computer GmbH
fax. : +49 89 7936 0039                            Ludwig Thoma Weg 11a
andrew@IFComputer.isar.de                          D-82065 Baierbrunn
