JBoss has announced the release of JBoss Drools 4.0 today. The declarative rule language and Eclipse Rule Flow modeler enable concise and powerful rules with a number of authoring approaches, the Rule Engine provides for high performance execution and the Business Rules Management System brings enterprise class management. JBoss Drools 4.0 can be summarised as:

• More expressiveness.
• More powerful declarative keywords.
• Hibernate ready, with 'from' keyword for evaluating external data.
• Pluggable dialects, with new MVEL dialect.
• New Rule Flow and Eclipse modeller.
• Better Performance.
• IDE Improvements.
• Enterprise Ready with Web 2.0 Business Rules Management Studio.

Resources:

• Presentation from Skills Matter
• What's new in JBoss Rules 4.0

congrats on the hard work. peter

For up to date information on Drools and Expert Systems please do subscribe to our blog at: http://blog.athico.com Also here is a more detailed enhancements list: Language Expressiveness Enhancements * New Conditional Elements: from(hibernate ready), collect, accumulate and forall * New Field Constraint operators: not matches, not contains, in, not in, memberOf, not memberOf * New Implicit Self Reference field: this * Full support to Conditional Elements nesting, for First Order Logic completeness. * Support to multi-restrictions and constraint connectives && and || * Parser improvements to remove previous language limitations, like character escaping and keyword conflicts * Support to pluggable dialects and built-in support to Java and MVEL * Complete rewrite of DSL engine, allowing for full l10n * Fact attributes auto-vivification for return value restrictions and inline-eval constraints * Support to nested accessors, property navigation and simplified collection, arrays and maps syntax * Improved support to XML rules * Experimental Clips parser support Core Engine Enhancements * Native support to primitive types, avoiding constant autoboxing * Transparent optional Shadow Facts * Rete Network performance improvements for complex rules * Rule-Flow * Stateful and Stateless working memories (rule engine sessions) * Support for Asynchronous Working Memory actions * Rules Engine Agent for hot deployment and BRMS integration * Pluggeable dialects and and full support to MVEL scripting language * Dynamic salience for rules conflict resolution * Parameterized Queries * halt command * Sequential execution mode, faster performance and uses less memory * Pluggable global variable resolver IDE Enhancements * Support for rule break-points on debugging * WYSIWYG support to rule-flows * New guided editor for rules authoring * Upgrade to support all new engine features Business Rules Management System - BRMS * User friendly web interface with nice WEB 2.0 ajax features (GWT) * Package configuration * Rule Authoring easy to edit rules both with guided editor ( drop-down menus ) and text editor * Package compilation and deployment * Easy deployment with Rule Agent * Easy to organize with categories and search assets * Versioning enabled, you can easily replace yours assets with previously saved * JCR compliant rule assets repository Miscellaneous Enhancements * Slimmed down dependencies and smaller memory footprint Enjoy :) The Drools Team Mark Proctor, Michael Neale, Edson Tirelli, Kris Verlaenen, Fernando Meyer http://blog.athico.com

Congratulations and respect to the whole Drools team!

Congrats to the Drools team! Drools is an EXCELLENT product. I've been very impressed with its performance, reliability, and documentation. As was our client. Mark Proctor and the rest of the team have done a great job. Dan Diephouse

When used at enterprise level, a good rule management system is the most important.

I haven't had a chance to evaluate that part yet, but I do have to admit, their BRMS screenshots look pretty enticing.

The rest of the stuff is technically excellent, but it is the BRMS that will do most to drive enterprise adoption. Sad but true , pretty beats useful most of the time. Just as well the BRMS is a very good product. Paul (Link to wiki on integrating JBoss rules, Spring and Maven)

To those of us familiar with Prolog, there are alternatives - though the tool support seems far better for Drools / JBoss Rules. http://mandarax.sourceforge.net/ (Was it Drools that used to be an expensive commercial product?)

Drools 4.0 now has a Clips parser, if you happen to also like Lisp syntax, you can also use Clips-like syntax to author your rules. I believe Drools was an open source project from the very beginning. Backward reasoning is in Drools' roadmap by all means. Please have a look at Taseree project: http://blog.athico.com/search/label/search%20space It might be adopted/enhanced as Drools-Solver in Drools' next major release. The Drools team's vision is beyond a Rete-based rule engine. They are on their way to make Drools an AI platform which is accessible to programmers working with business clients. PS: Drools 4 has a nice rule scripting shell too. You can access the whole java API from the shell by saying "java org.mvel.MVELSH" if you have Drools installed. More information here: http://mvel.codehaus.org/

Drools 4.0 now has a Clips parser, if you happen to also like Lisp syntax, you can also use Clips-like syntax to author your rules.

I am creating a custom tool to design and display the "rules". The next step is to figure out how to get them from my structure to "Drools". I've no experience with Clips, but I will give it a look. Any pointers?

I am creating a custom tool to design and display the "rules". The next step is to figure out how to get them from my structure to "Drools". I've no experience with Clips, but I will give it a look. Any pointers?

Well you can go direct to the AST (which can be a bit fiddly, but that is what happens with the CLIPS and DRL and XML parsers - they go direct to the AST - all the classes that end in "Descr" are the AST). An alternative approach is to map to your own specific AST (using a parser generator of your choice - we recommend antlr) and then either map that to the drools AST, or use a Visitor to walk the tree and dump out clp or drl. All the above have different pros and cons depending on how coupled you want it to be to the drools-compiler module (in this case coupling is not as bad a thing as it sounds, as a lot of the hard work is done by the code that deals with the drools AST).

Congrats to the team. One of the ehancement is "Better Performance". Is there a benchmark that shows the performance increases ? Thanks.

I've tested past alpha releases. My experience is version 4 is 5x faster on most cases. for Manners benchmark, the performance improve is huge. With version 3.0, it used to take about 90seconds with manners 128 on an old AMD Sempron 1.8ghz with 1.5Gb of RAM. Using an earlier alpha release from feb/march, the time dropped down to 16 seconds with the same hardware. that's roughly 5x improvement for a stress test. Keep in mind manners is a performance stress test and doesn't represent common use case. If a developer writes rules that way, it's probably by accident and not on purpose. manners benchmark is great for people who write rule engines. PR people tend to abuse benchmarks in general. There are other techniques for improving performance, but for 99.99% of the cases, the performance should be more than sufficient. Most people looking at business rule engines care more about rule management and things like rule repository, rule validation and ide. peter

Congratulations for a whole team of professionals. Thank you for your hard work.

We are still missing 3 main things: analytics ontology modelling testing Our next release will be quite quick, I'm hoping approx 3 months. That release should hopefully have analytics and testing in it, as I want to get those out as soon as possible. Ontology modelling will take a little longer so will be in release after that, along with prolog style backwards chaining (for a full hybrid engine) and Complex Event Processing(CEP)/Event Stream Processing(ESP). We have some community members working on interesting stuff too. There is a solver framework nearing completion, which will be badged drools-solver - this will help with scheduling and rostering style applications which have large search spaces. Someone else is working on a pluggeable belief system so you can annotate the constraints and the rules with a certainty value and have that evaluated by the subsystem (fuzzy logic, probability etc). We definitely want to see this progress from beyond a rules engine to an integrated AI platform for behavioural modelling. Mark http://blog.athico.com

We definitely want to see this progress from beyond a rules engine to an integrated AI platform for behavioural modelling.

That sounds great. I have been think about how to use it in that enviroment. I've not had much time to look into it yet. Most of the analytics is currently done with SQL and SAS.

We definitely want to see this progress from beyond a rules engine to an integrated AI platform for behavioural modelling.

That sounds great. I have been think about how to use it in that enviroment. I've not had much time to look into it yet. Most of the analytics is currently done with SQL and SAS.

I could be wrong, but the kind of analytics Mark is referring to is rule validation and rule analysis. to the best of my knowledge, no commercial product today provides impact analysis for rules, objects or data in the IDE or rule repository. Most provide rule validation, which is different than impact analysis. This is one area where the business rule industry on the whole is lacking. I've discussed some of this stuff with the drools developers, though I am not a member of the team. the kind of impact analysis I'm thinking are changes toe the object model, rule templates and facts. peter

We definitely want to see this progress from beyond a rules engine to an integrated AI platform for behavioural modelling.

That sounds great. I have been think about how to use it in that enviroment. I've not had much time to look into it yet. Most of the analytics is currently done with SQL and SAS.

I could be wrong, but the kind of analytics Mark is referring to is rule validation and rule analysis. to the best of my knowledge, no commercial product today provides impact analysis for rules, objects or data in the IDE or rule repository. Most provide rule validation, which is different than impact analysis. This is one area where the business rule industry on the whole is lacking.

I've discussed some of this stuff with the drools developers, though I am not a member of the team. the kind of impact analysis I'm thinking are changes toe the object model, rule templates and facts.

peter

I was thinking that is what he was referring to too. Until I read the last two sentences.

We definitely want to see this progress from beyond a rules engine to an integrated AI platform for behavioural modelling.

That sounds great. I have been think about how to use it in that enviroment. I've not had much time to look into it yet. Most of the analytics is currently done with SQL and SAS.

I could be wrong, but the kind of analytics Mark is referring to is rule validation and rule analysis. to the best of my knowledge, no commercial product today provides impact analysis for rules, objects or data in the IDE or rule repository. Most provide rule validation, which is different than impact analysis. This is one area where the business rule industry on the whole is lacking.

I've discussed some of this stuff with the drools developers, though I am not a member of the team. the kind of impact analysis I'm thinking are changes toe the object model, rule templates and facts.

peter

I was thinking that is what he was referring to too. Until I read the last two sentences.

Mark P - can you clarify? Thx.

The analytics will cover things like number ranges and change impact. Change impact will basically show you dependency graphs of rules, facts and fields and let you know what can potentially be impacted. We will also try and provide "hints" on things like recursion, conflicting logic and rule subsumption. We still have a lot of R&D to do and looking for more volunteers :) A student is doing some initial work for us, before we get started on this in earnest: http://rikkola.blogspot.com/ Mark

If you're interested in rule impact analysis, send me an email woolfel AT gmail DOT com. I have several entries on the topic on my blog. peter