[GHC] #6093: Kind polymorphism fails with recursive type definition using different kind

[GHC]

#6093: Kind polymorphism fails with recursive type definition using different kind
----------------------------------------+-----------------------------------
  Reporter:  Ashley Yakeley             |          Owner:                  
      Type:  bug                        |         Status:  closed          
  Priority:  normal                     |      Milestone:                  
 Component:  Compiler (Type checker)    |        Version:  7.6.1           
Resolution:  fixed                      |       Keywords:                  
        Os:  Unknown/Multiple           |   Architecture:  Unknown/Multiple
   Failure:  GHC rejects valid program  |     Difficulty:  Unknown         
  Testcase:  polykinds/T6093            |      Blockedby:                  
  Blocking:                             |        Related:                  
----------------------------------------+-----------------------------------

Comment(by goldfire):

 I would argue that the failure of the second of Ashley's recent examples
 is not a bug, according to the manual. See
 [http://www.haskell.org/ghc/docs/latest/html/users_guide/kind-
 polymorphism.html#id620778 this section]. Polymorphic recursion is only
 enabled for a datatype with a "complete user-supplied kind signature".
 According to the manual, such a datatype ''must'' be defined with GADT
 syntax and with a top-level {{{::}}}. The second example above does not
 have a top-level (i.e. outside of any parentheses) {{{::}}}, so it does
 not have a "complete user-supplied kind signature" and thus cannot use
 polymorphic recursion.

 As I understand it, Simon struggled a bit with the design around this
 issue, trying to balance kind inference with polymorphic recursion. The
 final result is something of a compromise, and is admittedly a little
 fiddly and can be unexpected behavior. However, it is well defined in the
 manual.

-- 
Ticket URL: <http://hackage.haskell.org/trac/ghc/ticket/6093#comment:13>
GHC <http://www.haskell.org/ghc/>
The Glasgow Haskell Compiler