[GHC] #12804: forever contains a space leak

[GHC]

#12804: forever contains a space leak
-------------------------------------+-------------------------------------
           Reporter:  tomjaguarpaw   |             Owner:
               Type:  bug            |            Status:  new
           Priority:  normal         |         Milestone:
          Component:                 |           Version:  8.0.1
  libraries/base                     |
           Keywords:                 |  Operating System:  Unknown/Multiple
       Architecture:                 |   Type of failure:  Runtime
  Unknown/Multiple                   |  performance bug
          Test Case:                 |        Blocked By:
           Blocking:                 |   Related Tickets:
Differential Rev(s):                 |         Wiki Page:
-------------------------------------+-------------------------------------
 4.9.0.0's `forever` (implemented in terms of `Applicative`) has a space
 leak for certain base `Monad`s.  The old `forever` (implemented in terms
 of `Monad`) does not.

 See these messages for details

 * https://mail.haskell.org/pipermail/haskell-cafe/2016-October/125177.html
 * https://mail.haskell.org/pipermail/haskell-cafe/2016-October/125178.html
 * https://mail.haskell.org/pipermail/haskell-
 cafe/2016-November/125443.html

 This is not necessarily `forever`'s fault.   It seems likely that the
 broken behaviour that occurs with `ReaderT` and `StateT` (on `IO`) could
 be fixed by a specialised implementation of `*>` (for `ReaderT` and
 `StateT`).

 Perhaps, then, this bug should ultimately be fixed in `transformers` (and
 various other packages which supply `Applicative`s) but it is a regression
 introduced by base-4.9.0.0 so I think it's worthwhile to discuss here.

--
Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/12804>
GHC <http://www.haskell.org/ghc/>
The Glasgow Haskell Compiler