There is, but you have to look at things slightly differently - and we have to really work on a new DRL to better support it, so it's all very experimental and bleeding edge. It's also probably more lambdas and closures than pure functions.
Where I see lambdas helping is in creating meta control systems for expert systems, attaching functions to events handlers. I'd quite like to see C# style delegator operators, for compact ways of assigning event handlers.
I'd then like to start exposing subsystems event models and life cycles in DRL, to have handlers added. These functions can have state (closures) and control how execution happens. I sketched a document some time ago on ideas going on in my head - it's very rough and much is already out of date on how I'd like to do it:
So for instance at the moment Drools has "activation-group", which provides hard coded behaviour. What if instead we exposed the RuleModule (group) and it had events you could attach to
onRuleAdded
onRuleRemoved
onBeforeRuleFired
onAfterRuleFired
I could add a function to "onAfterRuleFired" which would cancel all the other waiting rules. So I'd like to actually expose a RuleModule in DRL as declarable artefact (just a sketch, not working or compilable code)
module MyModule {
onAfterRuleFire() {
conflictSet.cancelAll();
}
}
While the above is declaration based approach, it could use an api approach too
getRuleModule( "MyModule" ).onAfterRuleFire += new RuleModuleEvent() {
conflictSet.cancelAll();
}
I'd more recently sketch out some ideas on controlling rule execution semantics - based on the DADO research paper for "all" instances to be fired. If you scroll down to the bottom, you can see a suggested life cycle for rule firings:
Other proposed life cycles:
- OnBeforeMatchFire
- Before a single Match instance is executed
- OnAfterMatchFire
- After a single Match instance is executed
- onBeforeAllFire
- Before any Matches in the set have fired
- onAfterAllFire
- After all Matches in the set have fired
- onDeleteMatch
- invoked when a Match is deleted (no longer true)
- onUpdateMatch
- invoked when a Match is updated
- onChangeMatch
- invoked when a Match is either updated or deleted
I'm also starting to prefer the C# "on" prefix notation for events.
I think as well instead of doing it initially in DRL, a lot can be done in pure Java with annotations used for wiring and discovery. We can replace or atleast more easily expose conflict handling and all the "group" stuff we have now. Note a lot of these events already exist, but they are more for observing. So it's all about re-looking at our events, and our sub systems, thinking about how they can be better exposed to create a meta system.
@RuleModule( "MyModule" ) // RuleModule can be optional and inferred from class name.
public class MyClassName {
@ConflictSet
private ConflictSet set; // Tells it to inject the conflict on instiation
public void onAfterRuleFired(Match match) {
set.canceAll()
}
}
So over all our longer term research is looking into a meta expert system - that looks at better ways to expose sub systems and their life cycles and using event handlers to allow the user to control the behaviour. I also think we'll need a sort of macro system, to make this easily re-usable.
If this is work you are interested in research, then please pop onto irc - we'll be happy to mentor.
Mark
Ciao everybody, I have a question I would like to submit to this community, given Drools is my preferred choice for Expert Systems :)
A quick background: I'm not an Expert Systems / Rule engines expert, but I'm a happy power user "of Rules" since my CS Engineering thesis, a paper published, and used it with happy results at work and hobby projects. Currently boarded on a new project with Rules+Processes at work, and still very happy, so happy, I'm using Drools again in a new hobby project. [semi-brag-mode off]
Now, in recent times, I cannot ignore what (apparently) is a strong renaissance of Functional Programming, specifically all this momentum which Scala and Java 8's Lambda's are getting.
The Question I'm asking myself: is there a specific, or few, areas where Functional Programming could actually benefit Expert Systems so much to actually spawn new paradigms of writing Rules? Or benefit in a new generation of the Rete algorithm?
Or actually, Functional Programming will spawn a completely different paradigm of system which will go beyond Expert Systems the way are designed today?
Ignore my rant if you feel it's too generic, but I really wanted to seize a chance to ask the question to this community :)
Thank you for your feedback,
Ciao,
Matteo
rules-users mailing list
rules-users@lists.jboss.org
https://lists.jboss.org/mailman/listinfo/rules-users