Do startnode and endnode have accessors? like getStartnode() and
getEndnode()
Re the second question, you can use regular java to constraint fields with
a predicate expression:
n2 : Node( value -> (getValue() < n1 * r.getMetric()))
the -> ([java]) construct lets you constrain with any java expression that
evals to a boolean. Would that meet your needs?
On Wed, 26 Apr 2006 08:58:57 -0400, Alberto Siena <[EMAIL PROTECTED]> wrote:
Hi all,
I'm currently evaluating the use of drools for my research purposes and
I have
two questions for you. I omit the java code because I think the problem
is not
on that side, but if needed I can post it.
1. I have a directed graph with two nodes (N1 and N2) and (possibily) a
relation
(R) between them. N1 has a value, let say, 1.0; what I'm trying to do is
a
propagation mechanism that, if N2.value < N1.value, and the relation
actually
exists between the nodes, then assigns N1.value to N2.value; otherwise,
all
remains unchanged.
I wrote the following rule:
rule "Propagate values"
when
n1 : Node( v1: value > 0 );
n2 : Node( value < v1 );
r : Relation();
eval( r.startnode == n1 );
eval( r.endnode == n2 );
then
n2.value = n1.value;
modify( n2 );
end
And it works. However, if I understood correctly, eval() is a second-best
choice. So I wrote this one:
rule "Propagate values"
when
n1 : Node( v1: value > 0 );
n2 : Node( value < v1 );
r : Relation( startnode == n1, endgoal == n2 );
then
n2.value = n1.value;
modify( n2 );
end
But it won't work (the condition is never satisfied). Could you please
explain
me why? (The objects in the workingMemory are the same, only the rule
changes)
2. Let suppose the relation has a "metric", let say 0.5. The meaning of
the
metric is this: if N2.value is less than (N1.value * R.metric), then
N2.value =
(N1.value * R.metric); otherwise, it remains unchanged.
The first problem is: the "*" operator is not supported in rules. So,
how could
I produce the same effect without using it? (And btw, will mathematical
operators be available in the future)?
Thank you
Alberto
--
Using Opera's revolutionary e-mail client: http://www.opera.com/mail/
