I would have thought it would have been sensible to base floating point maths on the IEEE spec, as thats what most hardware implements, and there are libraries to emulate it for everything else. Does haskell use IEEE primitives for things like sin/cos and sqrt? I am really surprised this area seems so neglected.
I for one would like to see floats and doubles following the IEEE spec, complete with constructors for +/- Infinity, NaN, and primitives.
Keean
Glynn Clements wrote:
MR K P SCHUPKE wrote:
Double already has +Inf and -Inf; it's just that Haskell doesn't have
(AFAIK) syntax to write them as constants.
In the source for the GHC libraries it uses 1/0 for +Infinity and -1/0 for -Infinity, so I assume these are the "official" way to do it.
Personally I would define nicer names:
positiveInfinity :: Double positiveInfinity = 1/0
negativeInfinity :: Double
negativeInfinity = -1/0
Or just:
infinity = 1/0
and use -infinity for the negative.
One other nit: isn't the read/show syntax for Haskell98 types supposed to valid Haskell syntax?
From http://www.haskell.org/onlinereport/derived.html#derived-text
The result of show is a syntactically correct Haskell expression containing only constants, given the fixity declarations in force at the point where the type is declared.
[Note: the above sentecne refers specifically to derived instances, but induction would require that it also holds for base types.]
However:
Prelude> let infinity = 1/0 :: Double Prelude> show infinity "Infinity" Prelude> read (show infinity) :: Double Infinity Prelude> Infinity <interactive>:1: Data constructor not in scope: `Infinity'
_______________________________________________ Haskell-Cafe mailing list [EMAIL PROTECTED] http://www.haskell.org/mailman/listinfo/haskell-cafe
