[GHC] #8995: When generalising, use levels rather than global tyvars

[GHC]

#8995: When generalising, use levels rather than global tyvars
-------------------------------------+------------------------------------
        Reporter:  simonpj           |            Owner:
            Type:  bug               |           Status:  new
        Priority:  normal            |        Milestone:
       Component:  Compiler          |          Version:  7.8.2
      Resolution:                    |         Keywords:
Operating System:  Unknown/Multiple  |     Architecture:  Unknown/Multiple
 Type of failure:  None/Unknown      |       Difficulty:  Unknown
       Test Case:                    |       Blocked By:
        Blocking:                    |  Related Tickets:
-------------------------------------+------------------------------------

Comment (by simonpj):

 I've pushed my work-in-progress to branch `wip/T8995-level-
 generalisation`.

 Validate says this
 {{{
 Unexpected failures:
    callarity/should_run   StrictLet [exit code non-0] (normal)
    codeGen/should_run     cgrun010 [exit code non-0] (normal)
    ghci.debugger/scripts  break026 [bad stdout] (ghci)
    ghci.debugger/scripts  print020 [bad stderr] (ghci)
    ghci/scripts           ghci046 [bad stderr] (ghci)
    perf/haddock           haddock.base [stat not good enough] (normal)
    polykinds              T6068 [bad stderr] (ghci)
 }}}
 Pretty good really.  A couple of these are Lint failures, though.

 The big reason I can't now get rid of `getGlobalTyVars` is the kind-
 generalisation point.

 Simon

--
Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/8995#comment:1>
GHC <http://www.haskell.org/ghc/>
The Glasgow Haskell Compiler