[GHC] #15147: Type checker plugin receives Wanteds that are not completely unflattened

[GHC]

#15147: Type checker plugin receives Wanteds that are not completely unflattened
-------------------------------------+-------------------------------------
        Reporter:  nfrisby           |                Owner:  (none)
            Type:  bug               |               Status:  new
        Priority:  normal            |            Milestone:  8.6.1
       Component:  Compiler (Type    |              Version:  8.4.1
  checker)                           |             Keywords:
      Resolution:                    |  TypeCheckerPlugins
Operating System:  Unknown/Multiple  |         Architecture:
                                     |  Unknown/Multiple
 Type of failure:  None/Unknown      |            Test Case:
      Blocked By:                    |             Blocking:
 Related Tickets:                    |  Differential Rev(s):
       Wiki Page:                    |
-------------------------------------+-------------------------------------

Comment (by simonpj):

 I've looked at why the plugin is seeing flattened givens.  It has been
 this way for some time.

 Currently, when we go "under" an implication constraint we flatten the
 Givens; and we only unflatten them when we've finished with that
 implication. When calling a plugin, we have Givens from a whole stack of
 enclosing implications, and none of them will be unflattened.

 See the `inert_fsks` field of `InertSet`; and `TcSMonad.unflattenGivens`
 which uses that field to guide unflattening.  You can see that
 `unflattenGivens` is called by `nestImplicTcS` which goes under an
 implication.

 It's quite awkward to unflatten them, in fact... even accessing the
 unflattening info for the stack of implications would require extra
 plumbing.

 Humph.  How bad would it be just to document the status quo?

-- 
Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/15147#comment:8>
GHC <http://www.haskell.org/ghc/>
The Glasgow Haskell Compiler