[GHC] #14880: GHC panic: updateRole

[GHC]

#14880: GHC panic: updateRole
-------------------------------------+-------------------------------------
        Reporter:  RyanGlScott       |                Owner:  goldfire
            Type:  bug               |               Status:  new
        Priority:  normal            |            Milestone:  8.8.1
       Component:  Compiler (Type    |              Version:  8.2.2
  checker)                           |
      Resolution:                    |             Keywords:  TypeInType
Operating System:  Unknown/Multiple  |         Architecture:
 Type of failure:  Compile-time      |  Unknown/Multiple
  crash or panic                     |            Test Case:
      Blocked By:                    |             Blocking:
 Related Tickets:  #15076            |  Differential Rev(s):  Phab:D4769,
       Wiki Page:                    |  Phab:D5141
-------------------------------------+-------------------------------------

Comment (by simonpj):

 Sigh.

 For a start, `closeOverKinds` stupidly goes via `FV`, which is silly in
 our new setup.  Perhaps
 {{{
 closeOverKinds tv_set = tv_set `unionVarSet` mapUnionVarSet kind_fvs
 tv_set
   where
     kind_fvs tcv = ty_co_vars_of_type (varType tv) emptyVarSet emptyVarSet
 }}}
 You might want to define
 {{{
 tyCoVarsOfType_unclosed :: Type -> TyCoVarSet
 tyCoVarsOfTYpe_unclosed ty = ty_co_var_of_type emptyVarSet emptyVarSet
 }}}

 Next, I'd worry about that `mapUnionVarSet`.  It seems that checking for
 dups,
 and avoiding unions, is a help.  So perhaps
 {{{
 closeOverKinds tv_set = close emptyVarSet tv_set
  where
    close :: TyCoVarSet -> TyCoVarSet -> TyCoVarSet
    -- (close acc tvs2) extends acc with tcvs closed over kinds
    -- Invariant: acc is already closed-over-kinds
    close acc tcvs = nonDetFoldUFM close1 acc tvs

    close1 :: TyCoVar -> TyCoVarSet -> TyCoVarSet
    close1 tcv acc
      = add1 (close kind_vars acc)
      where
        kind_vars = ty_co_vars_of_type (varType tcv) acc emptyVarSet
        add1 tvs2 | tcv `elemVarSet` tvs2 = tvs2
                  | otherwise             = extendVarSet tvs2 tcv
 }}}
 By using the the "in-scope set" (first argument) of `ty_co_var_of_type`,
 I'm saying "don't bother with variables whose kind we have already closed
 over".  And I'm trying to use `extendVarSet`, not `unionVarSet`.

 I'm not certain I have this right, but I think it's close.

-- 
Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/14880#comment:122>
GHC <http://www.haskell.org/ghc/>
The Glasgow Haskell Compiler