[GHC] #2059: Erroneous results in trigonometric functions for > double-precision values

[GHC]

#2059: Erroneous results in trigonometric functions for > double-precision values
----------------------------------+-----------------------------------------
    Reporter:  daniel.is.fischer  |        Owner:       
        Type:  feature request    |       Status:  new  
    Priority:  normal             |    Milestone:       
   Component:  Compiler           |      Version:  6.8.1
    Severity:  normal             |   Resolution:       
    Keywords:                     |     Testcase:       
Architecture:  x86                |           Os:  Linux
----------------------------------+-----------------------------------------
Changes (by ajd):

  * type:  bug => feature request
  * summary:  Erroneous results in trigonometric functions => Erroneous
              results in trigonometric functions for >
              double-precision values

Comment:

 The problem is that the Double type cannot store that much precision and
 so GHC can't calculate trig functions at all accurately. (This is actually
 documented, see [http://www.haskell.org/ghc/docs/latest/html/users_guide
 /wrong-compilee.html]. Since this is the way the compiler is documented to
 work, I'm changing the status to feature request.)

 From my understanding, in order to solve this, one would need to implement
 an arbitrary precision ''floating-point'' system in GHC (currently, we
 only have arbitrary precision ''integers'' through GMP). This ''could'' be
 done through something like MPFR, but I'm guessing that would be a huge
 job and potentially plagued with the same problems as GMP, as the latter
 is a dependency of the former.

-- 
Ticket URL: <http://hackage.haskell.org/trac/ghc/ticket/2059#comment:1>
GHC <http://www.haskell.org/ghc/>
The Glasgow Haskell Compiler