Datalog has an <https://github.com/racket/datalog/tree/master/tests|impressive test suite> however! :open_mouth:

If you implement one of the term indexing techniques would it improve the performance of datalog?

Probably not. Datalog has a number of constraints that I don’t have in my case: https://en.wikipedia.org/wiki/Datalog#Features,_limitations_and_extensions

I implemented an imperfect discrimination tree, and it turned out to be much simpler than I thought once the idea is understood

Next I need to improve it to a matching/perfect discrimination tree. (it’s a little more complicated as I need to track the substitution along the branches. Should be easy, but there’s room for mistakes :grin:)

For datalog, a hash is probably enough since literals are ‘flat’ and simple

> implemented an imperfect discrimination tree, > May I see ?

pm