[rules-users] Who writes your rules? Dev vs. BA

[Vincent Legendre]

My own feelings about that is yes, rules writer must know the objects 
they manipulate.
So, to answer your question, if the object model is very technical, Devs 
must write rules. If not, BA can do.

The main point is how much technical is your model objects. If you are 
starting a fresh new application, you should consider designing the 
model objects in order to make them as close as possitble to the BA's 
concepts, thus making possible (and natural) for BA to write rules directly.

My personnal experiences, not only with drools (exactly the same problem 
appears with commercial products too), is to try to provide model 
objects that "speak" to BA. This way I do not need a full complex DSL 
(which is quite heavy to maintain if you have to create a phrase for 
each "mapped" attribute of every object). To achive that, I often use a 
Facade that transform my real technical objects into more 
business-oriented objects used to feed the engine.

But you still need some training, even if you model objects match 
perfectly the concepts used, at least to explain what is a rule, how it 
is executed (to let them answer by their own "why it is looping ?" "Why 
this rule does not trigger?" ...)


Le 01/02/2011 17:08, cclark01 a écrit :
> My company is looking to use Guvnor in a new application we're writing.  Part
> of the appeal is the idea that Business Analysts will be able to go and
> write rules by themselves.  However, in doing a small proof-of-concept, it
> appears that you need intimate knowledge of the object model in order to
> write any meaningful rules.
>
> I've read up on DSL and it looks like that might help things, but it also
> sounds like DSL is the kind of thing that gets built up over time.
>
> So... what's your experience?  Do BA's or Developers write the rules in your
> applications?  Does the complexity of the object model play a role in who
> can write rules, or is it just a matter of training?
> If you've created a DSL, how easy/quick was it?  If not, any particular
> reason why not?
>
> I'm looking for some "real life" experiences so that we can have appropriate
> expectations going into the project.
>
> Thanks,
> Chris