9:11 a.m.
We have 5 rules that are firing out of salience order for a particular set
of 4 objects. The rules have a branch on the RHS which sometimes retracts
the A object. There are reasons why the branch is not in the LHS not shown
in this sample.
The 5 rules are like the following. The salience values
are 70, 65, 60, 55, 50
rule "a1"
salience 70
agenda-group "b"
when
a: A(name customevaluator "X")
B(name = "B")
then
// sometimes retract
end
There is one B object.
Another rule fires (at salience 0) and inserts 4 A objects.
1) A1 runs rule at salience 70 and is retracted
2) A2 runs rule at salience 70 and is not retracted
3) A3 runs rule at salience 70 and is retracted
4) A4 runs rule at salience 70 and is not retracted
5) A4 runs rule at salience 65 and is retracted
6) A2 runs rule at salience 50 and is retracted
I would have expected step 6 to have A2 run at salience 65 but instead it
went from 70 to 50 skipping 60 and 65. Is my understanding of the rule
rhs firing order wrong?
I'm using a 5.1 snapshot from Dec 23 2009.
I searched the JIRA and did not see anything related.
9:55 a.m.
Probably you have checked this, but just to make sure:
If you omit rule "a1": does fact A2 fire with rule "a2" at salience
65?
What does "...rules are like the following..." mean? I guess they are
not identical, that would be pointless. So, in which way do they differ?
-W
On Tue, Mar 30, 2010 at 4:11 PM, Jared Davis <sunray(a)davisprogramming.com>wrote:
We have 5 rules that are firing out of salience order for a
particular set
of 4 objects. The rules have a branch on the RHS which sometimes retracts
the A object. There are reasons why the branch is not in the LHS not shown
in this sample.
The 5 rules are like the following. The salience values
are 70, 65, 60, 55, 50
rule "a1"
salience 70
agenda-group "b"
when
a: A(name customevaluator "X")
B(name = "B")
then
// sometimes retract
end
There is one B object.
Another rule fires (at salience 0) and inserts 4 A objects.
1) A1 runs rule at salience 70 and is retracted
2) A2 runs rule at salience 70 and is not retracted
3) A3 runs rule at salience 70 and is retracted
4) A4 runs rule at salience 70 and is not retracted
5) A4 runs rule at salience 65 and is retracted
6) A2 runs rule at salience 50 and is retracted
I would have expected step 6 to have A2 run at salience 65 but instead it
went from 70 to 50 skipping 60 and 65. Is my understanding of the rule
rhs firing order wrong?
I'm using a 5.1 snapshot from Dec 23 2009.
I searched the JIRA and did not see anything related.
_______________________________________________
rules-users mailing list
rules-users(a)lists.jboss.org
https://lists.jboss.org/mailman/listinfo/rules-users
11:58 a.m.
I've narrowed it down to a smaller example. The order of insertion seems to
trump the rule salience.
The following rule inserts 4 objects. Inserted as A4,A3,A2,A1 the rules
sequence is unexpected. Inserted A1,A3,A4,A2 they fire as expected.
rule "setup.two"
when
then
// generates out of salience rule execution
insert (new RuleLink("A4", new Double(65), null));
insert (new RuleLink("A3", new Double(70), null));
insert (new RuleLink("A2", new Double(65), null));
insert (new RuleLink("A1", new Double(70), null));
/* // works ok
insert (new RuleLink("A1", new Double(70), null));
insert (new RuleLink("A3", new Double(70), null));
insert (new RuleLink("A4", new Double(65), null));
insert (new RuleLink("A2", new Double(65), null));
*/
end
Here are 3 of the 6 identical LHS rules differing only by salience
(70,65,60,55,50,0):
rule "a70"
salience 70
when
link: RuleLink( linkName: name matches "A.", d :
aasDouble)
RuleLink( name == "SKINS", partsO : a)
then
System.out.println("START 070 " + d);
if ( d > 69.9) {
System.out.println(drools.getRule().getName() + " retract " +
linkName);
retract(link);
} else {
System.out.println(drools.getRule().getName() + " no retract " +
linkName);
}
end
rule "a65"
salience 65
when
link: RuleLink( linkName: name matches "A.", d :
aasDouble)
RuleLink( name == "SKINS", partsO : a)
then
System.out.println("START 065 " + d);
if ( d > 64.9) {
System.out.println(drools.getRule().getName() + " retract " +
linkName);
retract(link);
} else {
System.out.println(drools.getRule().getName() + " no retract " +
linkName);
}
end
rule "a60"
salience 60
when
link: RuleLink( linkName: name matches "A.", d :
aasDouble)
RuleLink( name == "SKINS", partsO : a)
then
System.out.println("******* ERROR ************");
System.out.println("START 060 " + d);
if ( d > 59.9) {
System.out.println(drools.getRule().getName() + " retract " +
linkName);
retract(link);
} else {
System.out.println(drools.getRule().getName() + " no retract " +
linkName);
}
end
// One more setup rule
rule "setup.one"
when
then
List skinParts = new ArrayList();
insert (new RuleLink("SKINS", skinParts, null));
end
Here is the test run:
START 070 70.0
a70 retract A1
START 070 65.0
a70 no retract A2
START 070 70.0
a70 retract A3
START 070 65.0
a70 no retract A4
START 065 65.0
a65 retract A2
******** ERROR ********
START 060 65.0
a60 retract A4
A4 never ran rule 065. It skipped to rule 060 from 070.
The error is very sensitive to other rules in the package.
Is this the expected behavior?
(I would like to keep the reasons for a branch on the RHS vs an eval on the
LHS out of this discussion.)
--
View this message in context:
http://n3.nabble.com/User-and-or-Code-problem-with-insert-retract-and-sal...
Sent from the Drools - User mailing list archive at Nabble.com.
5385
days inactive
5385
days old
2 comments
2 participants
participants (2)
-
Jared Davis -
Wolfgang Laun