#2182: ghc sessions (--make, --interactive, ghc api) erroneously retain instances -------------------------------------+------------------------------------- Reporter: claus | Owner: ezyang Type: bug | Status: new Priority: normal | Milestone: 7.10.1 Component: Compiler | Version: 6.9 (Type checker) | Keywords: Resolution: | Architecture: Unknown/Multiple Operating System: | Difficulty: Unknown Unknown/Multiple | Blocked By: Type of failure: GHC | Related Tickets: accepts invalid program | Test Case: | Blocking: | Differential Revisions: | -------------------------------------+------------------------------------- Changes (by ezyang): * cc: simonpj (added) * owner: => ezyang Comment: Plan changed again! Now the idea is to maintain the set of orphan modules reachable from DIRECTLY imported modules, instead of the set of interface files. This ALMOST works. The big problem is when we mkLocalInstance, we need to know (immediately) if it's an orphan or not, because that entry in the instance environment may become enshrined forever if we have a GHCi session involved. I think we can pull this off by moving the logic from MkIface for calculating if an instance is an orphan to the typechecker. -- Ticket URL: http://ghc.haskell.org/trac/ghc/ticket/2182#comment:9 GHC http://www.haskell.org/ghc/ The Glasgow Haskell Compiler