So @oli -- this is a pretty old proposal -- any thoughts on it now?
I'm not sure without testing that we'll get actual benefits from this until there are more consumers of
MIR. I think we can close it until we see a concrete use case
I think I agree, although I do want to see this happen
I was hoping to start leading the way towards extracting the type system into a library, maybe that's a good pre-req regardless
for example, maybe we want a MIR library to actually just reference a type library, and be generic over the
I: Interner and not over types?
that would at least be an easier first step than what I had in mind