🚀 go-pugleaf

In article <6555@spool.cs.wisc.edu> g-suders@rocky.CS.WISC.EDU (S. Sudarshan) writes:
>I posted this article with distribution: us several days back, but haven't
>got any replies.
That posting never reached here.

There was a collection of real programs collected at the Prolog Benchmarking
Workshop;  I'll let the people at Aerospace answer that part of this message.

>As a course project this semester, I'm doing a study of "real life" prolog
>programs.  The aim of this study is to find the extent to which "real life"
>prolog programs are purely declarative (and don't depend on the control 
>strategy) and the extent to which the control strategy dependency is fairly
>small.

I can answer this one for you right away:  0.01%.  Good Prolog programmers
try to keep their code pure; even so cuts are, um, not terribly rare.  And
a lot of real life Prolog programmers aren still learning how to use the
language effectively.

>This ultimate aim is to study how useful bottom up evaluation strategies
>would be, and to see what changes could be made to these to mimic some 
>(simple) forms of control.

If this is the question you are really interested in, the idea of surveying
existing code seems back to front.  Before Prolog existed, there was no
body of logic programs just waiting for an efficient implementation.  You
won't get people writing programs which would run well on a bottom-up
system until there is a bottom-up system for them to run their programs on.

Do you know about the NAIL! project?  That's a bottom-up logic programming
system aimed at data-baseish applications.