[GHC] #14425: approxRational is wrong when (rat - eps) overflows in negative direction

[GHC]

#14425: approxRational is wrong when (rat - eps) overflows in negative direction
-------------------------------------+-------------------------------------
        Reporter:  danielk           |                Owner:  (none)
            Type:  bug               |               Status:  new
        Priority:  normal            |            Milestone:
       Component:  libraries/base    |              Version:  8.2.1
      Resolution:                    |             Keywords:
Operating System:  Unknown/Multiple  |         Architecture:
 Type of failure:  Incorrect result  |  Unknown/Multiple
  at runtime                         |            Test Case:
      Blocked By:                    |             Blocking:
 Related Tickets:                    |  Differential Rev(s):
       Wiki Page:                    |
-------------------------------------+-------------------------------------

Comment (by danielk):

 I suggest the following simple fix. Convert rat and eps to Rational before
 subtracting and adding epsilon. This eliminates the possibility of
 overflow (in either direction).


 {{{
 approxRational          :: (RealFrac a) => a -> a -> Rational
 approxRational rat eps  =  simplest (toRational rat - toRational eps)
 (toRational rat + toRational eps)
 ...
 }}}

-- 
Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/14425#comment:1>
GHC <http://www.haskell.org/ghc/>
The Glasgow Haskell Compiler