Rule engine development experience sharing - reddit

background . I work in a team , Responsible for managing a bunch of different rule engines . One of them is pure Java noodles , Only the people who developed it understand , The other is in a proprietary BRMS Running on the system , Tens of thousands of rules , Another one is developed by me , Is a decision table system .

When building a rule engine , The most important thing you can do is limit its functionality . In your case , You should have at least three separate rule engines ： An account logic , A check on product availability , Another problem with finding customer accounts . This will help limit possible spaghetti .

The second most important thing you can do is make sure that every rule can be tested 、 restructure , And you can guarantee this set of rules . So , I recommend avoiding the use of separate languages that enforce rules . It can be powerful , But they can be test nightmares , And specialized rule languages are usually not as good tools as mainstream languages . If you use something like a decision table , It's much easier to validate the input and convert it correctly into classes in your own code .

Last , I recommend that you prevent any form of status or ordering in your rules . If you need to stay in shape between rounds of execution , You're probably actually talking about different types of rules , for example , Establish different rule sets to determine whether customers can place orders （ Boolean output ）, And another rule set to determine the amount of credit they receive （ Decimal output ）, Instead of putting them together , use 0 To indicate that the customer cannot place an order .