[GHC] #8518: panic ghci when probably using type families incorrectly

10 Nov 2013

      #8518: panic ghci when probably using type families incorrectly
------------------------------+-------------------------------------
       Reporter:  HanStolpo   |             Owner:
           Type:  bug         |            Status:  new
       Priority:  normal      |         Milestone:
      Component:  Compiler    |           Version:  7.6.3
       Keywords:  ghci panic  |  Operating System:  Unknown/Multiple
   Architecture:  x86         |   Type of failure:  GHCi crash
     Difficulty:  Unknown     |         Test Case:
     Blocked By:              |          Blocking:
Related Tickets:              |
------------------------------+-------------------------------------
 Hi I am a beginner and was playing around with type families when ghc gave
 me a panic message and suggested I log a bug.

 I am running Windows 7 with GHC 7.6.3 installed using Haskell Platform
 2013.2.0.0.

 Here is the reduced code producing the panic:

 {{{
 {-# LANGUAGE TypeFamilies #-}
 import Data.Maybe
 import Control.Applicative

 class Continuation c where
     type Z c
     type B c
     type F c
     continue ::  c -> (Z c) -> (B c) -> Maybe ((F c), c)

 callCont :: Continuation c => c -> (Z c) -> (B c) -> Maybe (F c)
 callCont c z b = rpt (4 :: Int) c z b
     where
         rpt 0 c' z' b' = fromJust (fst <$> (continue c' z' b'))
         rpt i c' z' b' = let c'' = fromJust (snd <$> (continue c' z' b'))
 in rpt (i-1) c''

 main = putStrLn ""
 }}}

 Loading the code in WinGHCi and calling main results in the following

 {{{
 GHCi, version 7.6.3: http://www.haskell.org/ghc/  :? for help
 Loading package ghc-prim ... linking ... done.
 Loading package integer-gmp ... linking ... done.
 Loading package base ... linking ... done.
 Prelude> :cd C:\Dev
 Prelude> :load "Crash.hs"
 [1 of 1] Compiling Main             ( Crash.hs, interpreted )
 Ok, modules loaded: Main.
 *Main> main
 ghc: panic! (the 'impossible' happened)
   (GHC version 7.6.3 for i386-unknown-mingw32):
         nameModule
 <<details unavailable>>

 Please report this as a GHC bug:  http://www.haskell.org/ghc/reportabug

 *Main>
 }}}

 There is obviously something degenerate with the code but I guess the
 compiler should give a better error message.

 Note it actually compiles, i.e. if the file itself is just compiled but
 not linked in it does not give any errors.

 If the degenerate function 'callCont' is changed as follows.
 {{{
 {-# LANGUAGE TypeFamilies #-}
 import Data.Maybe
 import Control.Applicative

 class Continuation c where
     type Z c
     type B c
     type F c
     continue ::  c -> (Z c) -> (B c) -> Maybe ((F c), c)

 callCont :: Continuation c => c -> (Z c) -> (B c) -> Maybe (F c)
 callCont c z b = rpt (4 :: Int) c z b
     where
         rpt i c' z' b' = fromJust (fst <$> (continue c' z' b'))

 main = putStrLn ""
 }}}

 Then you get the following compile error:
 {{{
 Crash.hs|12 col 18 error| Occurs check: cannot construct the infinite
 type: uf0 = Maybe uf0
 || Expected type: Maybe (F c)
 ||   Actual type: F c
 || In the return type of a call of `rpt'
 || In the expression: rpt (4 :: Int) c z b
 || In an equation for `callCont':
 ||     callCont c z b
 ||       = rpt (4 :: Int) c z b
 ||       where
 ||           rpt i c' z' b' = fromJust (fst <$> (continue c' z' b'))
 }}}

--
Ticket URL: http://ghc.haskell.org/trac/ghc/ticket/8518
GHC http://www.haskell.org/ghc/
The Glasgow Haskell Compiler