On Tuesday, 30 April 2019 12:42:14 UTC-4, Stephan Merz wrote:
I don't have any experience with expert systems, but I guess in principle what you suggest should be possible. I am a bit afraid, though, about TLC being overwhelmed with the state space, and I'd consider using constraint solving for finding values that match guards of different rules and then explore the actions of these rules.
I have the same worries about state space (even if I were to do some "smart" things to limit it), and I've also been thinking that constraint solving might be the right approach. Thanks for helping give confidence to my thoughts!