#11120: Missing type representations -------------------------------------+------------------------------------- Reporter: goldfire | Owner: Type: bug | Status: patch Priority: high | Milestone: 8.0.1 Component: Compiler | Version: 7.11 Resolution: | Keywords: Operating System: Unknown/Multiple | Architecture: | Unknown/Multiple Type of failure: None/Unknown | Test Case: Blocked By: | Blocking: Related Tickets: | Differential Rev(s): Phab:D1769 Wiki Page: | -------------------------------------+------------------------------------- Comment (by bgamari): So I have a very rough cut of Richard's proposal. The primary wrinkle that I ran into here is that you are forced to produce the `GHC.Prim` representations as `CoreExpr`s in order to wire them in (as wired-in identifiers are just identifiers without a definition but instead a compulsory unfolding, which is Core). This is a bit unfortunate as in the usual case we produce representations as standard `HsExpr`s, meaning we must duplicate the code for producing type representations. There are three options I can see here, 1. Accept the code duplication and move on with life 2. Use compulsory unfoldings for all type representations, allowing us to drop the current `HsExpr` logic in `TcTypeable` 3. Instead make the `GHC.Prim` representations merely known-key and inject the bindings into some other module (like `GHC.Types`, since `GHC.Prim` doesn't have associated object code). This means we'd keep some of the special-case that we currently have, but on the other hand we could drop the hand-written type representations. -- Ticket URL: http://ghc.haskell.org/trac/ghc/ticket/11120#comment:16 GHC http://www.haskell.org/ghc/ The Glasgow Haskell Compiler