My question is around the performance impact of using declarative rules. While I understand the basic nuances like using backward chaining can at times be beneficial over forward chained expressions and so forth, my question is really around:
1. What is recommended from a performance standpoint - for developer to performance calculations manually or leave it to a declare rule? - especially in scenarios where a property is set only once in the entire application (say from a result of a decision tree) and can be set either using an activity/data transform or using a declare rule - which option to choose and why?
2. How much declare rules to use if an application is calculation intensive?
3. Any concrete proof regarding the above?
4. Any reading material already present regarding the performance impact of the inference engine?
It would depend on whether the property is set more often than it is read.
Since you said the property is set only once, you would want to use the flavor of declarative that computes the dependent property values when that first property is set, rather than computing the first property's value every time it is read. /Eric