[GHC] #11339: Possible type-checker regression in GHC 8.0

[GHC]

#11339: Possible type-checker regression in GHC 8.0
-------------------------------------+-------------------------------------
        Reporter:  hvr               |                Owner:  simonpj
            Type:  bug               |               Status:  closed
        Priority:  highest           |            Milestone:  8.0.2
       Component:  Compiler (Type    |              Version:  8.0.1-rc1
  checker)                           |
      Resolution:  fixed             |             Keywords:
Operating System:  Unknown/Multiple  |         Architecture:
                                     |  Unknown/Multiple
 Type of failure:  None/Unknown      |            Test Case:
                                     |  typecheck/should_compile/T11339,
                                     |  T11339b, T11339c, T11339d; partial-
                                     |  sigs/should_compile/T11339a
      Blocked By:                    |             Blocking:
 Related Tickets:                    |  Differential Rev(s):
       Wiki Page:                    |
-------------------------------------+-------------------------------------
Changes (by simonpj):

 * status:  new => closed
 * testcase:   =>
     typecheck/should_compile/T11339, T11339b, T11339c, T11339d; partial-
     sigs/should_compile/T11339a
 * resolution:   => fixed


Comment:

 This is a complicated ticket, and I've spent an unreasonably long time on
 the patch.  But I'm very happy with the result.

 There are a number of separate strands

 * comment:12, I think `fanout` definitely needs a type signature for `go`.
 Look at the GADT pattern match on `Refl` in the `Await` branch... GADTs
 always need a type signature!  I have no idea how 7.10 compiled this
 program; maybe a bug!

 * Difference between these two bindings:
 {{{
     Bazaar { getBazaar = t } = left sell s

     t = getBazaar (left sell s)
 }}}
   Well, the first falls under the Monomorphism Restriction, while the
   second does not.  Simple as that.

 * comment:4, item (2).  I disagree:
 {{{
 foo :: Show a => a -> String
 (foo) = show
 }}}
   Here the binding falls under the monomorphism restriction, and
   cannot be rescued by a type signature. Test:
 `typecheck/should_compile/T11339`
 {{{
     • Overloaded signature conflicts with monomorphism restriction
         t :: forall (f :: * -> *). Applicative f => (a -> f b) -> f t
 }}}
   Adding `-XNoMonomorhismRestriction` makes it fine.
   Test: `typecheck/should_compile/T11339b`

 * comment:4, item (5), yes this was a bad bug
 {{{
 foo :: _
 (foo) = id
 }}}
   Fixed: test is `partial-sigs/should_compile/T11339a`

 Tricky stuff.

--
Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/11339#comment:14>
GHC <http://www.haskell.org/ghc/>
The Glasgow Haskell Compiler