[rules-users] Declarative fact model or Java?

[Davide Sottara]

In alternative, if you want total decoupling, you can define /interfaces/
that your classes extend, and write the rules against the interfaces, rather
than the classes. This might help when you have different formats and
improves the portability of your rule base. 
The classes would have to implement interfaces statically, so it can be done
conveniently if you need to create your fact classes anyway. The drawback is
that you can't to that so easily if you don't have access to the source
code. We provided "traits" to add interfaces (a) to individual instances
rather than whole classes, or (b) to attach an interface to a class which
does not implement it.
One bit which is missing, now that I realize, is the possibility to have
declared classes implement DECLARED*** interfaces statically (anyone willing
to open a JIRA request? I'm feeling lazy right now :) )
D.


[Beware: arguable and debatable content below:]
Remark : In a perfect world, the rules should just process "logical facts",
i.e. data stored within pure beans. Behavior (methods) other
assertions/retractions/updates should be provided in a controlled way,
either using different, dedicated facts or globals. If this level of
"purity" can be attained, among other things, debugging and rule analysis
becomes much more feasible.



--
View this message in context: http://drools.46999.n3.nabble.com/rules-users-Declarative-fact-model-or-Java-tp3675001p3676227.html
Sent from the Drools: User forum mailing list archive at Nabble.com.