[GHC] #12791: Superclass methods could be more aggressively specialised.

[GHC]

#12791: Superclass methods could be more aggressively specialised.
-------------------------------------+-------------------------------------
        Reporter:  mpickering        |                Owner:
            Type:  bug               |               Status:  new
        Priority:  normal            |            Milestone:
       Component:  Compiler          |              Version:  8.0.1
      Resolution:                    |             Keywords:
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 ryantrinkle):

 simonpj: Yes.  I've always encouraged the abstract style (

 {{{
 (MonadReader r m, MonadWriter w m) => m a
 }}}

 rather than

 {{{
 ReaderT r (WriterT w m) a
 }}}

 ), so that the concrete monad transformer stack can be changed without
 modifying user code.  It has worked really well to help keep code clean
 and organized (especially with GHC 8.0's warnings for unused constraints).

 So, in a typical reflex widget, there will be several constraints, e.g.
 DomBuilder t m, PerformEvent t m, PostBuild t m.  Reflex is currently a
 superclass of all of these, since the 't' parameter is only meaningful for
 its Reflex instance.  I haven't looked into whether it's possible to
 eliminate Reflex from all of these classes, but I'll certainly try to do
 that if it'll help with optimization.

--
Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/12791#comment:14>
GHC <http://www.haskell.org/ghc/>
The Glasgow Haskell Compiler