[GHC] #14570: Untouchable error arises from type equality, but not equivalent program with fundeps

[GHC]

#14570: Untouchable error arises from type equality, but not equivalent program
with fundeps
-------------------------------------+-------------------------------------
        Reporter:  lexi.lambda       |                Owner:  (none)
            Type:  bug               |               Status:  new
        Priority:  normal            |            Milestone:
       Component:  Compiler (Type    |              Version:  8.0.2
  checker)                           |
      Resolution:                    |             Keywords:
Operating System:  Unknown/Multiple  |         Architecture:
 Type of failure:  GHC rejects       |  Unknown/Multiple
  valid program                      |            Test Case:
      Blocked By:                    |             Blocking:
 Related Tickets:                    |  Differential Rev(s):
       Wiki Page:                    |
-------------------------------------+-------------------------------------

Comment (by AntC):

 Replying to [comment:3 dfeuer]:
 > At the least, this is a rather poor error message.

 I'm wondering if the o.p. gave the full story about the error message. I
 retried the Type Family version, and got `f0 is untouchable` rejection,
 but against the type for `f`. The message went on to give more info `f is
 a rigid type variable ...`; plus a suggestion to `AllowAmbiguousTypes`.

 Switching that on does indeed suppress the message against `f`; then it
 just reappears against `g`, as per the o.p., but with less helpful info.
 (I've put more narrative against the StackOverflow question, link above.)

 Simon's explanation at comment:1 tells what's going on, but not really
 why. The Schrijvers et al paper didn't really help: it's aiming to explain
 FunDep inference in terms of Type Families/System FC, so doesn't tell
 why/how ghc's behaviour is different in this example.

 Re the `rigid type variable` message, there are some reasonable
 explanations on StackOverflow. Re `f0 is untouchable` not so much.

-- 
Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/14570#comment:4>
GHC <http://www.haskell.org/ghc/>
The Glasgow Haskell Compiler