[GHC] #13114: UniqSet definition seems shady

[GHC]

#13114: UniqSet definition seems shady
-------------------------------------+-------------------------------------
           Reporter:  dfeuer         |             Owner:  dfeuer
               Type:  task           |            Status:  new
           Priority:  low            |         Milestone:  8.4.1
          Component:  Compiler       |           Version:  8.1
           Keywords:                 |  Operating System:  Unknown/Multiple
       Architecture:                 |   Type of failure:  None/Unknown
  Unknown/Multiple                   |
          Test Case:                 |        Blocked By:
           Blocking:                 |   Related Tickets:
Differential Rev(s):                 |         Wiki Page:
-------------------------------------+-------------------------------------
 Currently,

 {{{#!hs
 type UniqSet a = UniqFM a
 }}}

 The key invariant of `UniqSet` is expressed in the somewhat-poorly-named
 `Note [Unsound mapUniqSet]`, and not enforced by the types. It seems
 likely that the clean thing is

 {{{#!hs
 newtype UniqSet a = US (UniqFM a)
 }}}

 Unfortunately there's an awful lot of code using `UniqSet` and assuming
 it's the same as `UniqFM`. To make this work, we'd need to expand the
 `UniqSet` API somewhat and figure out what to do at use sites using it
 interchangeably with `UniqFM`.

--
Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/13114>
GHC <http://www.haskell.org/ghc/>
The Glasgow Haskell Compiler