[GHC] #13804: MonoLocalBinds/RankNTypes type inference regression in GHC 8.2

[GHC]

#13804: MonoLocalBinds/RankNTypes type inference regression in GHC 8.2
-------------------------------------+-------------------------------------
        Reporter:  RyanGlScott       |                Owner:  (none)
            Type:  bug               |               Status:  new
        Priority:  highest           |            Milestone:
       Component:  Compiler (Type    |              Version:  8.2.1-rc2
  checker)                           |
      Resolution:                    |             Keywords:
Operating System:  Unknown/Multiple  |         Architecture:
 Type of failure:  GHC rejects       |  Unknown/Multiple
  valid program                      |            Test Case:
      Blocked By:                    |             Blocking:
 Related Tickets:  #11698            |  Differential Rev(s):
       Wiki Page:                    |
-------------------------------------+-------------------------------------

Comment (by simonpj):

 I see what is happening here.  We have
 {{{
 f periph = let sr :: forall a. [a] -> [a]
                sr = ... periph ...

                sr' = sr
            in
            ...
 }}}
 Can we do type inference for `sr'`?  Of course!  Its only free var is `sr`
 which has a closed type.

 But from the point of view of static pointers, `sr'` mentions `sr` and
 `sr` mentions `periph`, so they could not be floated out.  So they aren't
 closed in that sense.

 We were mixing the two up.

 Fix incoming.

--
Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/13804#comment:4>
GHC <http://www.haskell.org/ghc/>
The Glasgow Haskell Compiler