[GHC] #9554: Pathological type family turns type error into runtime loop (was: Pathological type family seems to violate FC operational semantics)

[GHC]

#9554: Pathological type family turns type error into runtime loop
-------------------------------------+-------------------------------------
              Reporter:  goldfire    |            Owner:
                  Type:  bug         |           Status:  new
              Priority:  normal      |        Milestone:
             Component:  Compiler    |          Version:  7.8.2
            Resolution:              |         Keywords:
      Operating System:              |     Architecture:  Unknown/Multiple
  Unknown/Multiple                   |       Difficulty:  Unknown
       Type of failure:              |       Blocked By:
  None/Unknown                       |  Related Tickets:
             Test Case:              |
              Blocking:              |
Differential Revisions:              |
-------------------------------------+-------------------------------------

Comment (by goldfire):

 Another look at this revealed it's not as bad as I thought: the `Show`
 instance for `Proxy` is ''lazy'' in its argument. So, `foo` was never
 being evaluated.

 But, consider this:

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

 import Data.Proxy

 type family F a where
   F a = F (F a)

 foo :: Proxy (F Bool) -> Proxy (F Int)
 foo x = x

 main = case foo Proxy of Proxy -> putStrLn "Made it!"
 }}}

 This ''does'' properly force `foo`, and the result is a runtime loop. The
 core is similar to that in the original post, in that it contains clearly
 looping coercions.

 This is bad: it means that `-XUndecidableInstances` allows programs to go
 wrong, contrary to my (and many others') assumptions about
 `-XUndecidableInstances`.

--
Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/9554#comment:1>
GHC <http://www.haskell.org/ghc/>
The Glasgow Haskell Compiler