My first encounter with a rules engine was Teamplate -- way back in 2003 -- which we used to deliver a .NET-based authorization for expenditure (AFE) system for a large oil and gas services company in Visual Basic and C#.
Similar to purchase orders, AFEs are used for very large capital expenditures and involve many decision-makers from more than one company. Complex workflows, involving delegation, confidentiality, and hard limits for budgetary approval are dynamically determined based on rules, time limits, delegation, and individual approval limits.
If we had tried to code all that with if/then logic, we would probably still be trying to debug it. ;)
As your excellent article demonstrates, rules engines can be extremely useful and powerful solutions.
But requiring them to be Python-based casts far too narrow a net in my opinion. And letting developers implement libraries according to their often misguided understanding of what constitutes a rule can lead to problems. Also, source code is too opaque for most business people to understand and maintain themselves.
I'm always looking for something standards-based, which gives me confidence that a lot of smart, experienced subject matter experts have had a lot of heated arguments and boring meetings and reached consensus over a period of years, the result of which has then been battle-tested in products for years.
In the intervening years, I've discovered the Decision Model and Notation (DMN) and one of its most successful implementations in the Camunda BPM Platform (links below).
These visual and executable business rule (or "decision") models (as well as their BPMN business process model counterparts and the rules and workflow engine that runs them) are something I definitely recommend checking out.
