[GHC] #9429: Alternative to type family Any

[GHC]

#9429: Alternative to type family Any
-------------------------------------+-------------------------------------
              Reporter:  mboes       |            Owner:
                  Type:  feature     |           Status:  new
  request                            |        Milestone:
              Priority:  normal      |          Version:  7.9
             Component:  Compiler    |         Keywords:
            Resolution:              |     Architecture:  Unknown/Multiple
      Operating System:              |       Difficulty:  Unknown
  Unknown/Multiple                   |       Blocked By:
       Type of failure:              |  Related Tickets:  9097, 9380
  None/Unknown                       |
             Test Case:              |
              Blocking:              |
Differential Revisions:              |
-------------------------------------+-------------------------------------

Comment (by mboes):

 To be clear, as I have said before, I'm not arguing for reverting back to
 the status quo that we have in GHC 7.8. It's pretty clear that we want to
 avoid having (an infinite number of) exotic inhabitants of kinds that are
 intended to be closed (and often with a finite number of inhabitants).

 As I see it if the old `Any`-as-datatype is revived under a different
 name, without imposing any restrictions on what kinds it inhabits (i.e.
 only open kinds), it would have to be in an `Unsafe.*` module, and only
 brought in scope in some very clearly delineated modules. At least in the
 case of Cloud Haskell, there is no reason why the user needs to even know
 of its existence - it's use can be contained, "behind the scenes". That
 may well still be unsatisfactory.

 simonpj: I'm all for a better solution that is not a hack like
 `rank1dynamic` is, but do you have any ideas for how generating `TypeRep`s
 for polymorphic types might work? `rank1dynamic` has the merit of existing
 today, and the reason we're seeking to extend it is precisely to help with
 #7015 without having to rewrite some of the internals of `distributed-
 process`. Of course, if we do understand how to get type reps for
 polymorphic types, Facundo and I are happy to volunteer some time to do
 the rewriting.

 So again, there are 4 possible solutions here. I don't know which one
 would be best, but it would be nice to support a better story for Cloud
 Haskell, which already has a number of users, while still keeping to a
 minimal impact on the compiler.

--
Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/9429#comment:28>
GHC <http://www.haskell.org/ghc/>
The Glasgow Haskell Compiler