[GHC] #12465: Evil idea: Allow empty record field update syntax for types.

5 Aug 2016

      #12465: Evil idea: Allow empty record field update syntax for types.
-------------------------------------+-------------------------------------
           Reporter:  Iceland_jack   |             Owner:
               Type:  feature        |            Status:  new
  request                            |
           Priority:  normal         |         Milestone:
          Component:  Compiler       |           Version:  8.0.1
           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:
-------------------------------------+-------------------------------------
 Use case: Visible type application.

 I often work with constructors with many arguments, then I want to
 instantiate type variables with them:

 {{{#!hs
 show @(Vector _ _)
   :: Show a => Vector n a -> String

 -- Type arguments reversed, see
 https://github.com/ekmett/distributive/pull/18
 cotraverse @_ @(Vector _)
   :: (NATTY n, Functor f) => (f a -> b) -> (f (Vector n a) -> Vector n a')

 fmap @(Bazaar _ _ _)
   :: (t -> t') -> (Bazaar p a b t -> Bazaar p a b t')

 show @(Magma _ _ _ _)
   :: (Show a, Show i) => Magma i t b a -> String
 }}}

 This makes no sense since types have nothing to do with record or updates
 thereof but that syntax is often used to avoid parentheses and unneeded
 wildcard arguments:

 {{{#!hs
 isBar (Bar _ _ _ _ _) = True

 -- ===>

 isBar Bar{} = True
 }}}

 This also means you don't need to worry about the kind of your constructor
 wrt the type (class) variable it instantiates.

 ----

 Thus the proposal is to allow:

 {{{#!hs
 show @Vector{}
   :: Show a => Vector n a -> String

 cotraverse @_ @Vector{}
   :: (NATTY n, Functor f) => (f a -> b) -> (f (Vector n a) -> Vector n a')

 fmap @Bazaar{}
   :: (t -> t') -> (Bazaar p a b t -> Bazaar p a b t')

 show @Magma{}
   :: (Show a, Show i) => Magma i t b a -> String
 }}}

 ----

 This would save quite a few keystrokes in my daily coding but more
 importantly it saves cognitive load but I expect skepticism, especially
 since GHC would have to determine whether to interpret `show @Vector{}` to
 `show @Vector`, `show @(Vector _n)` or `show @(Vector _n _a)`.

--
Ticket URL: http://ghc.haskell.org/trac/ghc/ticket/12465
GHC http://www.haskell.org/ghc/
The Glasgow Haskell Compiler