[rules-users] Best pattern for getting results OUT of Integration server?

[Wolfgang Laun]

On 07/01/2014, dunnlow <dunnlow at yahoo.com> wrote:
> I have some confusion about running rules in the Integration Server (v5.5).
>
> I am sending POJOs to the server from a camel (v2.10) Spring client app.
> The client inserts a POJO (via batchexecutioncommand) and the rules change
> an attribute on the POJO to return the value to the client.
>
> I know, I know, changing attributes of facts is frowned upon (but when I
> started I didn't know any better... I swear).

Where did you get this notion, "changing...frowned upon"? It's a
natural enough operation, and the Command lets you request that the
fact object is returned to sender.

>
> The problem I'm having (surprise) is that updating the fact (to return to
> the client app) is causing unwanted activations, no-loops, lock-on-actives
> etc etc.

Consider calling setX() without modify/update. The returned object
should show the new values, but the Engine will be "blind" to the
updates (which, apparently, is just what you want).

>   So, I'm looking into changing this and inserting two string
> objects into my command so that the rules only modify those string objects.
>
> My questions:
> 1) Is that the best (/a good) way to implement this?

No, see below.

>
> 2) My big question/misunderstanding:  Does the integration server handle
> the
> batchexecutioncommands in series or is it inserting into the working memory
> in parallel (ie. are all executions from the first insert completed before
> the next batchexecutioninsert command is evaluated)?

A batch executions command is executed in a sequential way.

>
> For example,
>
> If I have a loop that inserts one POJO (Fact.class) and two POJOs (S.class,
> a simple class with one string class variable)  can I safely do something
> like below (is the engine able to keep Fact1 with its inserted S1 and S2
> objects separate from the next S objects that might be inserted immediately
> after OR might all Facts and S objects get confused (due to
> multi-threading,
> parallel operation, etc):
>
>       rule "evaluate a Fact POJO"
>       when
>           $fact1 : Fact (value > 100, card=="NONE")
>           $s1 : S ()
>           $s2 : S (this != $s1)
>       then
>            $s1 = "OK"
>            $s2 = "BLUECARD"

This will most certainly NOT do what I think that you think it does,
and it will even do twice, what it does, which doesn't make it any
better.


>       end
>
>       rule "remove inserted objects so memory doesn't fill up"
>           salience -1000
>       when
>           $completedFact : Fact ()
>       then
>           retract ($completedFact)
>       end
>
>       rule "remove inserted objects so memory doesn't fill up"
>           salience -1000
>       when
>           $completedS : S ()
>       then
>            retract($completedS)
>       end
>
> My client will use the resulting s1 and s2 values so I obviously don't want
> the returned s1 and s2 values for fact1 to be confused with the s1 and s2
> values for fact2.
>
> 3) Currently, I need to retract the pojos I insert once the processing is
> done(like above) -- kinda ugly.

That's the way it is - provided the session remains active and continues.

Another technique that's recommended for returning results is by means
of a global. (I think this is discussed in the Expert manual.)

-W

>  Is there a way to implement a "logical"
> insert for the integration server or a better way to handle this?
>
> Thanks for any thoughts!
> -J
>
>
>
> --
> View this message in context:
> http://drools.46999.n3.nabble.com/Best-pattern-for-getting-results-OUT-of-Integration-server-tp4027546.html
> Sent from the Drools: User forum mailing list archive at Nabble.com.
> _______________________________________________
> rules-users mailing list
> rules-users at lists.jboss.org
> https://lists.jboss.org/mailman/listinfo/rules-users
>