Hi folks,

I quite like the clean separation provided by maintaining a declarative fact model rather than a Java model. In particular to me, it seems to make it easier to manage rules in Guvnor separate from my runtime environment. However I can also see how maintaining a Java model simplifies some activities, avoiding having to map between and external Java model and an internal representation, and making it easy to modify facts from within a knowledge session.

Is there anybody out there who has experience doing a number of projects both ways, and has developed an opinion on whether one is preferable the other? Or is there perhaps more of a decision tree pertaining to which is better in particular situations?

And on a side-note, is there anything out there that helps with mapping between Java beans and declarative facts? I was wondering about hacking up some AOP aspects to enable me to annotate a Java model in a slightly JPA/ORM style to help with mapping/marshalling between the two.

Thanks for any thoughts.

Steve

stephen.masters@me.com