🚀 go-pugleaf

In article <536@dcl-csvax.comp.lancs.ac.uk>
simon@comp.lancs.ac.uk (Simon Brooke) provides some information about
Prolog systems for VAX VMS.  I'd like to correct some of it.

(1) If 'Edinburgh' Prolog is still the same system that its designer
    called NIP, it is a compiler.  A key feature of the original design
    was that there was to be no distinction between compiled and interpreted
    code:  clause/2, retract/1, listing/1 and so on work by running the
    compiled code in a special mode (ALS also use this technique; I do not
    know whether they were aware of the Clocksin/Byrd/Bowen work).

(2) PopLog does in fact compile Prolog all the way to actual VAX instructions.

(3) InterFace Prolog does have a C interface.  I don't particularly like it,
    but they've had it for years.

(4) The speed figures make Quintus Prolog look reasonably good, but I have
    to warn you that naive reverse is almost totally useless as a predictor
    of performance on real applications.  The ability of a system to keep
    its working set small can be of greater consequence.

(5) The VMS version of Quintus Prolog has the same foreign interface as
    the UNIX version, plus there are a couple of extra VMS-specific functions
    provided by Quintus Prolog to user C code.  It's all in the "System-
    Dependent Features Manual -- VMS Version".

(6) Quintus has a UNIX interface to Unify and Oracle.  Potential VMS
    customers: please let us know which data-base systems you would prefer
    a VMS interface to (RDB? whatever?) so that we can plan which one to
    do first.

(7) A general observation: anyone considering buying Quintus Prolog can
    get a 30-day evaluation copy (in Europe, check with AI Ltd) and find
    out exactly what it's like and how it performs.  I imagine that other
    Prolog companies make a similar offer.  There is no substitute for
    actually trying a product on the actual hardware and operating system
    version you intend to use.