[GHC] #9980: TcS monad is too heavy

[GHC]

#9980: TcS monad is too heavy
-------------------------------------+-------------------------------------
        Reporter:  simonpj           |                Owner:  (none)
            Type:  bug               |               Status:  new
        Priority:  normal            |            Milestone:
       Component:  Compiler          |              Version:  7.8.4
      Resolution:                    |             Keywords:
                                     |  TypeCheckerPlugins
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 darchon):

 I had no need for `unsafeTcPluginTcM` until GHC 8.5, and only because:

 * The evidence for `KnownNat` is a `Natural`, even though Core doesn't
 have "first-class" support for Natural literals (but if I understand
 correctly, somebody is working on improving this situation)
 * Evidence is no longer a separate data type, but just a normal core
 expressions (except for evidence for `Typeable`).

 So the causes for my need `unsafeTcPluginTcM` can simply be classified as
 a "series of unfortunate/unforeseen events". However, isn't that exactly
 the reason for `unsafeTcPluginTcM`s existence in the first place? To have
 an escape hatch for situations that weren't easily foreseeable.

-- 
Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/9980#comment:8>
GHC <http://www.haskell.org/ghc/>
The Glasgow Haskell Compiler