[rules-users] Dynamic Fact Types

[Wolfgang Laun]

(Fact Templates are over and out.)

You create a TypeDeclaration by writing a DRL declare, which results in a
considerable number of objects being created and available at run time:
TypeDeclaration, ClassDefinition, FactField and FieldDeclaration and several
other objects, the "accessors" So, you could create a piece of DRL text
containing "declare X ... end" and pass this to the builder and add the
resulting package to your Knowledge Base.

And then?

You'll need some code to get objects of type X into Working Memory, which
you can do using the reflection-like methods of FactType and FactField, but
this code still has to know about "X" and its field names.

OK, so you have some Xs in WM. And now?

You'll have to write some rules referring to X and its fields, and how can
you do that? More DRL text?

True, rule patterns can use interface names as type names, but the DRL
language doesn't let you declare a type with an "implements" clause, and
therefore the Engine simply wouldn't know that your X implements anything.

Finally, a poster on this list has stated just a few days ago that declared
types are considerably slower than Java types. This hasn't been confirmed or
explained by one of the developers (although the additional layers mentioned
previously are certainly not speeding things up).

I don't know enough about your problem to suggest anything. It could be that
sticking with the Object/JavaBeans model and generating and compiling Java
and DRL is best. Or it could be that an entirely different ontology model is
advisable (even though the Drools fact model is clearly JavaBean-ish). Or
maybe you shouldn't use rules at all ;-)

-W


On 9 February 2011 23:02, jwillans2 <james.willans at ceteva.com> wrote:

>
> Hello,
>
> I am looking for an approach to defining fact types
> dynamically/programmatically.  Having hunted about, I can see a few ways of
> achieving this.  My preferred approach to doing this would be Fact
> Templates
> as described in the blog article here:
>
> http://orangemile.blogspot.com/2008/07/drools-fact-template-example.html
>
> I can see two problems with this approaches.  Firstly, it seems to be an
> experimental feature and generally not supported.  Secondly, it only seems
> to work with primitive fields (perhaps I'm wrong here?).  I get the
> impression that the preferred option is to use Type Declarations.  In the
> documentation for Type Declarations, it suggests that the rules evaluate
> against an interface rather than an actual Java object and that the
> interface can be implemented using different methods (POJO, reflection ..).
> Is there any examples or documentation on how to do this beyond generated
> Java bean class? Just to be clear, I am looking to define the fact types
> programmatic in a similar manner to that described in the above blog link,
> and I am eager to avoid a compilation/class loader step.
>
> Many thanks for any help that can be offered,
>
> James
> --
> View this message in context:
> http://drools-java-rules-engine.46999.n3.nabble.com/Dynamic-Fact-Types-tp2462037p2462037.html
> Sent from the Drools - User mailing list archive at Nabble.com.
> _______________________________________________
> rules-users mailing list
> rules-users at lists.jboss.org
> https://lists.jboss.org/mailman/listinfo/rules-users
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.jboss.org/pipermail/rules-users/attachments/20110210/2840f187/attachment.html