Comments on: Martin Fowler dislikes rules engines. I’m not so sure

By: Joseph Gutierrez

Joseph Gutierrez — Mon, 12 Jan 2009 05:31:43 +0000

You might want to check out Ayende’s approach.

http://ayende.com/Blog/archive/2008/12/30/discussion-oo-101-solutions-and-the-open-close-principle-at.aspx

This is done in C#.Net

By: Frank

Frank — Sun, 11 Jan 2009 23:38:46 +0000

I see what you mean. It’s entirely possible that I am reading into his words what is not really there.

However, I still feel that there is a distinction between traditional rules engines (which are mostly what you will encounter as you pursue investigation into RETE) and a lighter-weight, more modern, multi-threaded approach.

When I went through a similar research process a few years ago it took me quite along time to cut through the mystique of RETE to find much concrete detail. What I found was nothing very special. A technique for caching and sharing partial evaluations. Something modern distributed systems do all the time.

By: Joseph Gutierrez

Joseph Gutierrez — Sun, 11 Jan 2009 20:53:55 +0000

My take on Fowler’s bliki is not that he is against a rules engine but that there are some inherent liabilities, and that when using a rules engine that you have to be mindful of those liabilities.

I also postulate that the same liabilities and benefits engendered from using a rules engine are also similar to using an event driven design.

My post on this is:

http://uglylispcode.wordpress.com/2009/01/09/rules-engine-or-event-collaboration/#more-184

Nothing published yet, but I’m researching RETE algorithm and it has connected me to inference engines and complex event processing.

There is a lot of meat in just this one subject.

IMHO this directly relates to “machine control”, which if you look into Peter Norvig you will find it fits up into his alley with GOOGLE.
Just starting a search on wikipedia