#1797: type equality test leads to a looping type checker
-------------------------+--------------------------------------------------
Reporter: guest | Owner:
Type: bug | Status: new
Priority: normal | Milestone:
Component: Compiler | Version: 6.8
Severity: normal | Resolution:
Keywords: | Difficulty: Unknown
Os: Unknown | Testcase:
Architecture: x86 |
-------------------------+--------------------------------------------------
Old description:
> The following code (similar to one of HList’s generic type equality
> tests) causes GHC to loop forever (or at least very long) with
> GHC 6.8.0.20071020 on i386-unknown-mingw32 and GHC 6.8.0.20071019 on i386
> -unknown-linux:
> {{{
> {-# LANGUAGE
> MultiParamTypeClasses,
> FunctionalDependencies,
> UndecidableInstances,
> OverlappingInstances,
> FlexibleInstances,
> EmptyDataDecls #-}
>
> data True
>
> data False
>
> class TypeEq type1 type2 result | type1 type2 -> result where
> typeEq :: type1 -> type2 -> result
>
> instance TypeEq soleType soleType True where
> typeEq _ _ = undefined
>
> instance (TypeCast False result) => TypeEq type1 type2 result where
> typeEq _ _ = undefined
>
> class TypeCast type1 type2 | type1 -> type2, type2 -> type1
>
> instance TypeCast soleType soleType
> }}}
New description:
The following code (similar to one of HList’s generic type equality tests)
causes GHC to loop forever (or at least very long) with GHC 6.8.0.20071020
on i386-unknown-mingw32 and GHC 6.8.0.20071019 on i386-unknown-linux:
{{{
{-# LANGUAGE
MultiParamTypeClasses,
FunctionalDependencies,
UndecidableInstances,
OverlappingInstances,
FlexibleInstances,
EmptyDataDecls #-}
data True
data False
class TypeEq type1 type2 result | type1 type2 -> result where
typeEq :: type1 -> type2 -> result
instance TypeEq soleType soleType True where
typeEq _ _ = undefined
instance (TypeCast False result) => TypeEq type1 type2 result where
typeEq _ _ = undefined
class TypeCast type1 type2 | type1 -> type2, type2 -> type1
instance TypeCast soleType soleType
}}}
Comment (by simonpj):
Definitely a bug; similar to #1795.
Question: all instances of this bug seem to be for programs that should be
rejected. Are there any examples of divergence on programs that should be
accepted?
Meanwhile, will think about a fix. I agree this is a serious regression.
Simon
--
Ticket URL: <http://hackage.haskell.org/trac/ghc/ticket/1797#comment:1>
GHC <http://www.haskell.org/ghc/>
The Glasgow Haskell Compiler_______________________________________________
Glasgow-haskell-bugs mailing list
[email protected]
http://www.haskell.org/mailman/listinfo/glasgow-haskell-bugs
