#10675: GHC does not check the functional dependency consistency condition correctly -------------------------------------+------------------------------------- Reporter: simonpj | Owner: Type: bug | Status: new Priority: normal | Milestone: Component: Compiler | Version: 7.10.1 Resolution: | Keywords: Operating System: Unknown/Multiple | Architecture: | Unknown/Multiple Type of failure: None/Unknown | Test Case: Blocked By: | Blocking: Related Tickets: | Differential Revisions: -------------------------------------+------------------------------------- Comment (by goldfire): Simon says:
I'm not sure what the right thing here is.
I say that we should reject these programs. They're bogus! This might make some affected people complain, but the situation seems similar to what happened between 7.6 and 7.8 where the FD check was tightened (most recently). Some people's programs stopped working, but it was all for a good reason. The examples all look like an attempt to encode instance chains. GHC has hitherto failed to notice the overlap, but it's been there all along. Happily, each of these examples could straightforwardly (if verbosely and unpleasantly) be refactored to use closed type families. Or, if we ever get around to finishing the work for #8634, folks who want the old behavior can just call their functional dependency to be dysfunctional (which it really is) and get on with it. -- Ticket URL: http://ghc.haskell.org/trac/ghc/ticket/10675#comment:6 GHC http://www.haskell.org/ghc/ The Glasgow Haskell Compiler