Yes, sorry, it was a honest mistake in my part (late night coding
session :P), but the bug is still there when that cut is removed. And
yes, that was what Matthias meant.
Allow me to copy a part of the SWI Prolog documentation, if it may be of
any help:
?- help(!).
! [ISO]
Cut. Discard choice points of parent frame and frames created
after the parent frame. As of SWI-Prolog 3.3, the semantics of the
cut are compliant with the ISO standard. This implies that the
cut is transparent to ;/2, ->/2 and *->/2. Cuts appearing in the
condition part of ->/2 and *->/2 as well as in \+/1 are local to
the condition.
t1 :- (a, !, fail ; b). % cuts a/0 and t1/0
t2 :- (a -> b, ! ; c). % cuts b/0 and t2/0
t3 :- call((a, !, fail ; b)). % cuts a/0
t4 :- \+(a, !, fail ; b). % cuts a/0
The problem seems to be in the scope of the cut. Racklog cut seems to
cross %rel boundary whereas Prolog cut does not. Maybe Racklog has a
different concept of 'parent frame'? Maybe Racklog cut was meant to be
lexically scoped, but turns out to be dynamically scoped? Just a wild
guess though...
%not and %if-then-else might also be affected by this bug.
(FYI, if a reply is needed, expect it to be late as I'll be on holiday
until Aug 27.)
Thank you.
On Mon, 2012-08-13 at 16:58 -0400, Ryan Culpepper wrote:
> On 08/13/2012 04:41 PM, Marco Alberti wrote:
> > Erik,
> >
> > On Sun, 12 Aug 2012 12:32:53 +0700, Erik Dominikus wrote:
> >
> >
> >> (define %a
> >> (%rel (x)
> >> ((x) (%b x) !)
> >> ((x) (%c x)) ))
> >
> > the cut in the first clause is not there in the Prolog version; hence the
> > different behaviour.
>
> The problem exists even if you remove this cut. The other cut (in %b) is
> still behaving incorrectly.
>
> (I assume that's what Matthias meant in his recent response.)
>
> Ryan
>
> ____________________
> Racket Users list:
> http://lists.racket-lang.org/users
____________________
Racket Users list:
http://lists.racket-lang.org/users
