A simple experiment would show you that calling fireUntilHalt() again,
after an exception occurs (in some RHS code) will simply continue with
processing what's left on the agenda.
Of course, the RHS code that throws the exception has only executed
partially, which could easily leave your system in an inconsistent state.
In order to determine which rule and which fact(s), you'd have to
activate event handling and to log the "current activation" in the
exception handler surrounding fireUntilHalt().
-W
2010/9/1 tom ska <tiberium.linux(a)gmail.com>:
Hello,
I think you understood me. Your solution is to validate facts in the rule.
OK - I think this is a good idea, but... What if rules are created by no-IT
(no-Math-aware ;) ) people? I can imagine situation where they do not
validate this issue (div by 0). What then? If I send 1000 facts to ksession,
and 901 has value set to 0? I want my 900 correct facts back. How can I
know, how many have been processed?
P.S. I don't want to use globals to count this.
Thanks,
tom.
2010/8/31 Edson Tirelli <tirelli(a)post.com>
>
> I am not sure I understand your problem. The rule clearly should NOT
> match facts with value 0, as it will raise division by zero error all the
> time. So, the solution, in my mind is:
> rule "r1"
> when
> p : CTestObj( value != 0 )
> then
> p.setValue( 1 / p.getValue() );
> end
>
> If you need to give a different treatment to objects with value == 0,
> just add another rule for them.
> []s
> Edson
>
> 2010/8/31 tom ska <tiberium.linux(a)gmail.com>
>>
>> Hello,
>> I have a class and rule:
>>
>> class CTestObj
>> {
>> private double value;
>>
>> void setValue(double v) {
>> this.value = v;
>> }
>>
>> double getValue() {
>> return this.value;
>> }
>> }
>>
>> rule "r1"
>> when
>> p : CTestObj
>> then
>> p.setValue( 1 / p.getValue() );
>> end
>>
>> As you can see, in this rule, there is risk, to put into stateFulSession
>> object with value set 0. I can't validate this, because, this problem has a
>> variation where I put into stateFulSession object with set no 0, but the
>> rule will modify it...
>> So - I can't validate this = fact with value 0 is correct.
>>
>> But I process with stateFulSession many, many facts... For example 1000.
>> And, after 900 facts it occurs, that fact number 901 has value set on 0. In
>> effect I get an exception: org.drools.runtime.rule.ConsequenceException:
>> java.lang.ArithmeticException: / by zero. And ksession stops processing
>> facts. But, 900 facts are processed! I don't want to process them second
>> time.
>> And this is my question:
>>
>> How can I know, which fact is first no-processed fact? Which one stoped
>> fireAllRules method?
>> Beacuse I don't want to process them another time.
>>
>> Greetings :)
>> tom.
>>
>>
>>
>>
>>
>> _______________________________________________
>> rules-users mailing list
>> rules-users(a)lists.jboss.org
>>
https://lists.jboss.org/mailman/listinfo/rules-users
>>
>
>
>
> --
> Edson Tirelli
> JBoss Drools Core Development
> JBoss by Red Hat @
www.jboss.com
>
> _______________________________________________
> rules-users mailing list
> rules-users(a)lists.jboss.org
>
https://lists.jboss.org/mailman/listinfo/rules-users
>
_______________________________________________
rules-users mailing list
rules-users(a)lists.jboss.org
https://lists.jboss.org/mailman/listinfo/rules-users