Re: [Haskell-cafe] syntactic anti-unification of TH terms --- best practice?

Don't you need to do this translation anyway, because Pat and Exp use different constructors?

Yes, somewhere I have to say how to convert Pat to Exp.

Perhaps you can reduce some of the lookup cases to looking up just Pat and Exp (eg, looking up in a Body or Clause or Pat or Exp could always give antiunifiers in terms of either Pat or Exp, so you need only one type of Map)?

So you mean use a specific (M.Map [Exp] Exp) instead of a general (M.Map [a] a)? Yes, this would make sense. Then I only translate the value I want to lookup (if at all) and not the whole map, which I can pass to the accordant state transformer straight away. I'll try it out. Thanks, Martin