🚀 go-pugleaf

In article <1990Mar1.210219.5687@Neon.Stanford.EDU> feldman@Neon.Stanford.EDU (Todd J. Feldman) writes:
>
>Can someone please help me with a bug in the following Prolog fragment:
>
Maybe this problem is related to Feldman's:

In the book 'AI Through Prolog', Neil Rowe uses something like this:

(1)   doo :- agenda(X), assertz(agenda(next)), write(X), fail.
  
When this rule starts there is ONE fact 'agenda(start)' in the database,
so the first query agenda(X) succeeds with X=start. 
Now the problem with all of our Prolog interpreters (Prolog-2 from 
Expert Systems International Ltd, Xprolog, SBprolog) is that they
are so 'smart' as to see that with only ONE agenda() fact in the
database, there is no need to keep a backtrack-point there, so that
fact is interpreted as 

agenda(start) :- !.

And following assertz'ed agenda's are NOT found by backtracking, so
the output is just 'start no', while what was intended is:
'startnextnextnextnextnextnext....' (database full) as a result.

When you start with TWO agenda facts it works as intended (try
calling 'doo' a second time)

My question: Is there ANY prolog interpreter that does what
             I (and apparantly N. Rowe too) expect ?

BTW, the workaround is of course to always have two 'dummy' facts in
     the database for cases like this. (or is there a better way ?)
-- 
Jan Derriks                  |       AHA-TMF (H.T.S. 'Amsterdam'),
jand@maestro.htsa.aha.nl     |       Europaboulevard 23,
(or ..hp4nl!htsa!jand)       |       1079 PC Amsterdam,
phone: +31 20423827          |       the Netherlands.