nprofile1qyt8wumn8ghj7un9d3shjtnyd968gmewwp6kytcqyzf5l64n2xk9azdrt6c96nwshexx45v4sduj7cyakmw005afuu05jm4rrqe (nprofile…rrqe) These are all good points.
I think the real question for me is how this effects my term/type *syntax*. E.g. now that (concrete syntax identifiers) can refer either to an element of a type or a proper judgment, should I have two mutually recursive syntax trees that I elaborate into or just one? And should syn/chk return a sum type between them?
Perhaps they really should be distinguished in the concrete syntax for clarity...