Coq is either too immature or too strict. Writing a proof in coq is far too difficult for what it does.
I agree that we need a proof 'compiler' though.
I agree that we need a proof 'compiler' though.
What do you think are the specific ways in which a proof compiler would be different?