Data.Numbers.CrackNum

A library for formatting/analyzing FP and Integer values

Internal representation of a Floating-point numbers

data FP

Complete internal representation for a floating-point number

Constructors

FP

Fields

• intVal :: Integer

  The value as represented as a full Integer. Storage purposes only.

• prec :: Precision

  FP precision.

• sign :: Bool

  Sign. If True then negative, otherwise positive.

• stExpt :: Int

  The exponent as it is stored.

• bias :: Int

  The implicit bias of the exponent.

• expt :: Int

  The actual exponent.

• fracBits :: [Bool]

  Bits in the fractional part

• bitLayOut :: String

  Layout representation

• kind :: Kind

  Floating-point kind (i.e., value)

data Precision

Floating point precision

Constructors

HP

Half precision; 16 bits = 1 sign + 5 exponent + 10 mantissa

SP

Single precision; 32 bits = 1 sign + 8 exponent + 23 mantissa

DP

Double precision; 64 bits = 1 sign + 11 exponent + 52 mantissa

Instances

Eq Precision

Methods

(==) :: Precision -> Precision -> Bool

(/=) :: Precision -> Precision -> Bool

Show Precision

Methods

showsPrec :: Int -> Precision -> ShowS

show :: Precision -> String

showList :: [Precision] -> ShowS

data IPrecision

Integer/Word precision

Constructors

W8

8-bit unsigned (byte)

I8

8-bit signed

W16

16-bit unsigned (word)

I16

16-bit signed

W32

32-bit unsigned (double-word)

I32

32-bit signed

W64

64-bit unsigned (quad-word)

I64

64-bit signed

Instances

Eq IPrecision

Methods

(==) :: IPrecision -> IPrecision -> Bool

(/=) :: IPrecision -> IPrecision -> Bool

Show IPrecision

Show instance for integer-precisions

Methods

showsPrec :: Int -> IPrecision -> ShowS

show :: IPrecision -> String

showList :: [IPrecision] -> ShowS

data Kind

Kinds of floating point values

Constructors

Zero Bool

Zero: 0. If Bool is true, then this is -0; otherwise +0.

Infty Bool

Infinity: oo. If Bool is true, then this is -oo, otherwie +oo.

SNaN

The signaling-NaN.

QNaN

The quiet-NaN.

Denormal

Denormalized number, i.e., leading bit is not 1

Normal

Normal value.

Creating FP values

floatToFP :: Float -> FP

Turn a Haskell float to the internal detailed FP representation

doubleToFP :: Double -> FP

Turn a Haskell double to the internal detailed FP representation

stringToFP :: Precision -> String -> FP

Convert the given string to a IEEE number with the required precision

integerToFP :: Precision -> Integer -> FP

Crack a Haskell Integer value as the given precision floating value. The Integer should be the value corresponding to the bit-pattern as the float is laid out in memory according to the IEEE rules.

Displaying FP and Int/Word values

displayFP :: FP -> String

Display a Floating-point number in a nicely formatted way. (This function is also available through the Show instance for FP, but is provided here for symmetry with displayWord.)

displayWord :: IPrecision -> Integer -> String

Display a Integer (signed/unsigned) number in a nicely formatted way

Orphan instances

Show FP

Show instance for FP

Methods

showsPrec :: Int -> FP -> ShowS

show :: FP -> String

showList :: [FP] -> ShowS