[GHC] #13249: Default signature check can be quite onerous

[GHC]

#13249: Default signature check can be quite onerous
-------------------------------------+-------------------------------------
        Reporter:  bgamari           |                Owner:  (none)
            Type:  feature request   |               Status:  closed
        Priority:  normal            |            Milestone:  8.2.1
       Component:  Compiler (Type    |              Version:  8.1
  checker)                           |
      Resolution:  invalid           |             Keywords:
Operating System:  Unknown/Multiple  |         Architecture:
 Type of failure:  GHC rejects       |  Unknown/Multiple
  valid program                      |            Test Case:
      Blocked By:                    |             Blocking:
 Related Tickets:  #12918            |  Differential Rev(s):
       Wiki Page:                    |
-------------------------------------+-------------------------------------

Comment (by RyanGlScott):

 Sorry for not noticing this sooner, but I agree with Simon that the
 situation is not as dire as it seems.

 It should be noted that the correct default type signature could actually
 be made a little shorter:

 {{{#!hs
 default remaining :: (MonadTrans t, MonadGet n, m ~ t n, Remaining m ~
 Remaining n) => m (Remaining m)
 }}}

 That is, the `MonadGet n` constraint implies the `Monad n` constraint. I
 don't think four constraints is a unreasonable amount to ask for, given
 that we're using the "lift something into a `MonadTrans`" design pattern,
 which requires repeating the constraints (and associated type families)
 for the type that is lifted anyway.

 Also, to answer Iceland_jack's question: no, `t m (Remaining m)` would not
 be valid, both in the sense that GHC will reject it and in a semantic
 sense, as you're conflating two different `Monad`s. The first `m` in `t m
 (Remaining m)` represents the `Monad` you're lifted, while the second `m`
 is for the type which is a `MonadTrans` (i.e, the thing you're lifting
 into). Failing to distinguish between these two things has led to all
 sorts of problems in the past (see #12784), which is one the reasons we
 introduced this check in the first place.

--
Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/13249#comment:6>
GHC <http://www.haskell.org/ghc/>
The Glasgow Haskell Compiler