[rules-dev] Firing rules only if action causes updates to data?

Premkumar Stephen prem18 at gmail.com
Tue Feb 27 19:27:59 EST 2007


Peter,

Thanks for answering my post but I guess I was not clear ;

I know how to modify the rule using what you have stated below. However,
currently in our system, (using Ilog), the analysts write in English as how
I had shown earlier:

There are 4000 such rules and they would not like to go and modify them.
Secondly, even if they were willing to put the physical effort, this  (c !=
0) is not a required condition for the rule to fire and this english version
(which Ilog calls BAL or Business Action Language) needs to match whats in
our requirements tool (Doors).

Also, our rule actions in most cases, contain multiple actions. They would
not want to pollute the rule view.

So, anyways, Ilog lets me extract this out to a rule file called the Irl
file. Now, I need to go behind the scenes and try to modify this file to the
way I want it before deploying it in production.

So, in other words, I am looking for a solution that may do these things in
one of the ways below or achieve in some way that I have not thought of:

1) The drools api might have some flag that can be switched on; this flag
will look at the action(s); do an 'AND' on the negation of the action(s) and
'AND' it to the conditions.
This potentially could be triggered from the UI (where rules are in
pseudo-english). Thus in this case, the Drl would look exactly the same as
before but the rule itself is not fired if c = 0

2) There could be an api that modifies the drl file ( either hand-coded or
generated from the JBoss Rules UI ) to update the drl to implement the rule
as "rule x" as Peter Lin shows. (btw, the action I am looking for is rule x
and not rule y)

Thanks!!


On 2/27/07, Peter Lin <woolfel at gmail.com> wrote:
>
>
> simple
>
> rule x:
>
> if
>   a = 5
>   b = 3
>   c != 0
> then
>   c = 0
>
> keep in mind that using negation NOT means that condition does not exist.
> In other words, if I have this rule
>
> rule y:
> if
>   a = 5
>   b = 3
>   not c = 0
> then
>   c = 0
>
> rule y means a is 5, b is 3 and no c is zero.  Say i assert 100 objects to
> the working memory and 1 of them has c = 0. it means rule y would never
> fire, since there is a fact that satisfies "not c = 0".
>
> hope that helps
>
> peter
>
> On 2/27/07, Premkumar Stephen <prem18 at gmail.com> wrote:
>
> > Hi Dev team,
> >
> > Its been fascinating to watch the growth of this tool. Here is my
> > question:
> >
> > Rule x:
> > if
> >    a = 5
> >    b = 3
> > then
> >    c = 0
> >
> > Now, I have written this rule to run when a = 5 and b =3. But actually,
> > I would like it to ideally run when a = 5 and b = 3 and c != 0
> >
> > Now, does the second scenario lead to any savings. In Ilog, the second
> > scenario will result in the rule NOT being added to the agenda whereas it
> > would in the first.
> >
> > I cannot code the second scenario into the rule as the rule ( as seen in
> > UI) needs to match the requirements version ( in Doors) and they do not want
> > to add these to the requirements. I would just like to make the second
> > version for performance. We have noticed that less number of rules would
> > fire.
> >
> > Now, if a = 5, b = 3 and c = 0, is there a way Drools can look at it and
> > say, let me not fire this rule since there is no practical use anyways.
> > If not, what are the ways to automate not firing rules in scenarios
> > where action does NOT cause updates to data.
> >
> > Thanks!!
> >
> > _______________________________________________
> > rules-dev mailing list
> > rules-dev at lists.jboss.org
> > https://lists.jboss.org/mailman/listinfo/rules-dev
> >
> >
>
> _______________________________________________
> rules-dev mailing list
> rules-dev at lists.jboss.org
> https://lists.jboss.org/mailman/listinfo/rules-dev
>
>


-- 
Regards,
Prem
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.jboss.org/pipermail/rules-dev/attachments/20070227/e2e0a4a6/attachment.html 


More information about the rules-dev mailing list