Re: [GHC] #14260: Type family in instance signature confuses GHC

20 Sep 2017

      #14260: Type family in instance signature confuses GHC
-------------------------------------+-------------------------------------
        Reporter:  Iceland_jack      |                Owner:  (none)
            Type:  bug               |               Status:  new
        Priority:  normal            |            Milestone:
       Component:  Compiler          |              Version:  8.2.1
      Resolution:                    |             Keywords:
Operating System:  Unknown/Multiple  |         Architecture:
                                     |  Unknown/Multiple
 Type of failure:  None/Unknown      |            Test Case:
      Blocked By:                    |             Blocking:
 Related Tickets:                    |  Differential Rev(s):
       Wiki Page:                    |
-------------------------------------+-------------------------------------

Comment (by RyanGlScott):

 You //almost// pulled the wool over my eyes and made me think that GHC was
 infinitely looping without provocation. But then I realized that you
 conveniently left off some important language extensions needed to compile
 this code :)

 {{{#!hs
 {-# LANGUAGE TypeFamilies #-}
 {-# LANGUAGE UndecidableInstances #-}
 }}}

 The fact that you have to enable `UndecidableInstances` should be an
 important clue as to what's happening. You have these two type family
 instances:

 {{{#!hs
 type T () = T (W ())
 type T (W a) = T a
 }}}

 So if you have `u :: T (W ()) -> () -> (); u = undefined`, GHC attempts to
 generalize the type `T (W ()) -> () -> ()`, and thus reduce `T (W ())`.
 But that means:

 {{{
    T (W ())
 -> T ()
 -> T (W ())
 -> ...
 }}}

 Bam. Stack overflow. So I claim there is no bug here, only reckless use of
 `UndecidableInstances` and type families.

-- 
Ticket URL: http://ghc.haskell.org/trac/ghc/ticket/14260#comment:1
GHC http://www.haskell.org/ghc/
The Glasgow Haskell Compiler