[GHC] #12874: Read/Show Incompatibility in base

[GHC]

#12874: Read/Show Incompatibility in base
-------------------------------------+-------------------------------------
        Reporter:  andrewthad        |                Owner:  (none)
            Type:  bug               |               Status:  new
        Priority:  normal            |            Milestone:
       Component:  libraries/base    |              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:                    |
-------------------------------------+-------------------------------------
Changes (by RyanGlScott):

 * component:  Compiler => libraries/base


Comment:

 This does indeed seem to be a problem with `Proxy`'s `Read` instance. I've
 copied its current implementation below:

 {{{#!hs
 instance Read (Proxy s) where
   readsPrec d = readParen (d > 10) (\r -> [(Proxy, s) | ("Proxy",s) <- lex
 r ])
 }}}

 That `readParen (d > 10)` is a problem, since if the precendence `d` is
 greater than 10, then in order for parsing to succeed, `"Proxy"` must be
 surrounded by at least one set of parentheses. But, as you noted, this is
 too stringent of a requirement, since `show (Thing Proxy)` yields `"Thing
 Proxy"`, without any parentheses.

 One way to fix this is to ignore the precedence:

 {{{#!hs
 instance Read (Proxy s) where
   readsPrec _ = readParen False (\r -> [(Proxy, s) | ("Proxy",s) <- lex r
 ])
 }}}

 Then `"Thing Proxy"`, `"Thing (Proxy)"`, etc. all parse. But then again,
 this definition is basically equivalent to what `deriving Read` produces.
 Why not use that?

 FWIW, `Proxy` isn't the only type to suffer from this problem. It looks
 like `Coercion`, `:~:`, `:~~:`, and `U1` also have `Read` instances with
 the same issue.

--
Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/12874#comment:1>
GHC <http://www.haskell.org/ghc/>
The Glasgow Haskell Compiler