[rules-users] When is eval unavoidable?

[Bagwell, Allen F]

(Monday morning, back in the office.)

Ah, thanks!  I took out the final keywords and sure enough, it works.

Read the page link you provided. Good info there. I hadn't really delved into shadow facts much yet. Thanks for clearing this up!

:-)
-A

________________________________
From: rules-users-bounces at lists.jboss.org [mailto:rules-users-bounces at lists.jboss.org] On Behalf Of Edson Tirelli
Sent: Friday, April 25, 2008 1:11 PM
To: Rules Users List
Subject: Re: [rules-users] When is eval unavoidable?


    Allen,

    The problem is that you are making your methods "final". When you make your methods final, the shadow proxy can't intercept them to ensure working memory consistency. In your example, just removing the final modifier from your methods make them work as expected.

    If it is of any help, you can completely disable shadow facts by following the instructions in the manual or in this post:

http://blog.athico.com/2008/02/shadow-facts-what-you-always-wanted-to.html

    Also, the algorithm used in drools 5 will not require shadow facts anymore.

    The reason eval works in your example is that it bypass shadow facts.

    []s
    Edson


2008/4/25 Bagwell, Allen F <afbagwe at sandia.gov<mailto:afbagwe at sandia.gov>>:

I forgot to mention...

Drools 4.0.4 (the pre-compiled jars downloaded from the Drools web site)

Running on HP ProLiant/RedHat Enterprise 4.

-A

________________________________
From: rules-users-bounces at lists.jboss.org<mailto:rules-users-bounces at lists.jboss.org> [mailto:rules-users-bounces at lists.jboss.org<mailto:rules-users-bounces at lists.jboss.org>] On Behalf Of Edson Tirelli
Sent: Thursday, April 24, 2008 3:54 PM
To: Rules Users List
Subject: Re: [rules-users] When is eval unavoidable?


   Allen,

   Both your examples must work just fine:

$b : Y(running == true)

// attempting to find all applicants named Bob
$p : Person(firstName == "Bob")
$ap : Applicant(person == $p)

    I can't think about a reason for them to not work. If they aren't, can you please provide a self contained test case for us to evaluate?

    Answering your question, in Drools 4, assuming your facts are beans, there should be almost no reason for you to fall back into an eval. The usage of eval is usually related to the need for calculation inside a constraint or the call of functions.

   []s
   Edson




2008/4/24 Bagwell, Allen F <afbagwe at sandia.gov<mailto:afbagwe at sandia.gov>>:
I'm learning how to use Drools, but eval presents a problem to me when I am composing rules. I've read the manual and have a basic understanding of what it is for (very useful in certain situations!), but is there a better explanation of when it cannot be avoided?

For example when writing the LHS portion of a rule, I've run into a problem where literal restrictions alone are fine for one object but inadequate for another.  Like so:

$a : X(temp < 100)

$b : Y(running == true)

In both cases class X and class Y have standard Java bean set up for these fields with appropriate getters and setters, and the corresponding inserted facts see these fields being updated from time to time. But whereas the first fact causes rule activation when the temp field meets the rule critiera, the second will never work unless it is re-written as:

$b : Y()
eval ($b.isRunning() == true)


I've encountered a similar problem with in-line evals where Drools will accept an LHS like this:

// attempting to find all applicants named Bob
$p : Person(firstName == "Bob")
$ap : Applicant(person == $p)

but it will never cause activation unless you re-write it like this:

$p : Person(firstName == "Bob")
$ap : Applicant( eval(person == $p) )

despite the fact that the Applicant object's person field once set never changes.

So is there a more definitive explanation as to why one is forced to use eval when you would think that simple literal restrictions would be enough?  I've read in other posts that "eval is evil" and to best avoid it unless necessary, but this is perplexing me because I haven't yet discovered the way to think about rule composition that prevents eval use from appearing to be arbitrary. Mostly it's been write what I think *should* work and if it does then great, it if doesn't then keep adding evals until it does. For the record, I'm using the java dialect if that makes a difference.

Thanks!

-Allen





_______________________________________________
rules-users mailing list
rules-users at lists.jboss.org<mailto:rules-users at lists.jboss.org>
https://lists.jboss.org/mailman/listinfo/rules-users



--
Edson Tirelli
JBoss Drools Core Development
Office: +55 11 3529-6000
Mobile: +55 11 9287-5646
JBoss, a division of Red Hat @ www.jboss.com<http://www.jboss.com>

_______________________________________________
rules-users mailing list
rules-users at lists.jboss.org<mailto:rules-users at lists.jboss.org>
https://lists.jboss.org/mailman/listinfo/rules-users




--
Edson Tirelli
JBoss Drools Core Development
Office: +55 11 3529-6000
Mobile: +55 11 9287-5646
JBoss, a division of Red Hat @ www.jboss.com<http://www.jboss.com>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.jboss.org/pipermail/rules-users/attachments/20080428/257e7653/attachment.html