[GHC] #9151: Recursive default associated types don't kind-generalize properly

[GHC]

#9151: Recursive default associated types don't kind-generalize properly
-------------------------------------+------------------------------------
        Reporter:  goldfire          |            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 goldfire):

 Replying to [comment:4 archblob]:
 > > That second example is somewhat alarming -- the associated type family
 does not mention > any class parameters. I'm surprised that's accepted!
 >
 > @goldfire, wasn't this introduced in #7939 ?

 Perhaps, but if so, it was certainly not intentional. That work should
 make `b` poly-kinded, but it shouldn't necessarily make `F` accepted. I'm
 actually surprised that kind-checking strategies have much to do with this
 bug, but I haven't looked into it. When you change strategies, how does
 that affect other programs? As outlined in the very long comment
 explaining them and in #7939, kind-checking strategies are fiddly and
 subtle, a dangerous combination.

--
Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/9151#comment:6>
GHC <http://www.haskell.org/ghc/>
The Glasgow Haskell Compiler