FloatLayout structv0.1.0

The shape of one binary interchange format: where its fields are and what they mean.

pub struct FloatLayout {
    pub precision: uint;
    // private fields
}

Fields

Name	Type	Description
`precision`	`uint`	The significand precision, counting the leading bit whether or not it is stored.

Associated functions

`Extended80`

pub func Extended80() -> FloatLayout

The x87 extended format: sixty-four significand bits with the integer bit written out, in eighty bits.

Returns: the layout of x87's 80-bit extended format

Source

`Binary128`

pub func Binary128() -> FloatLayout

IEEE binary128: a hundred and thirteen bits of precision.

Returns: the layout of IEEE 754 binary128

Source

`Binary256`

pub func Binary256() -> FloatLayout

IEEE binary256: two hundred and thirty-seven bits of precision.

Returns: the layout of IEEE 754 binary256

Source

`Binary512`

pub func Binary512() -> FloatLayout

IEEE binary512: four hundred and eighty-nine bits of precision.

Returns: the layout of IEEE 754 binary512

Source

Methods

`SignificandFieldBits`

pub func SignificandFieldBits(self: &FloatLayout) -> uint

How many bits the stored significand field has: the precision, less the leading bit unless it is written.

Returns: how many bits the stored significand field occupies

Source

`MinimumExponent`

pub func MinimumExponent(self: &FloatLayout) -> int

The smallest exponent a normal value has, as the power of two its integer significand is scaled by.

Returns: the smallest exponent a normal value can have

Source

`MaximumExponent`

pub func MaximumExponent(self: &FloatLayout) -> int

The largest exponent a finite value has, on the same scale.

Returns: the largest exponent a finite value can have

Source

`DecimalDigits`

pub func DecimalDigits(self: &FloatLayout) -> uint

How many decimal digits the shortest exact rendering can need: one more than the precision is worth.

Returns: how many decimal digits round-trip through this format

Source

`WideClass`

pub enum WideClass: int32 {
    Zero = 0,
    Subnormal = 1,
    Normal = 2,
    Infinity = 3,
    NaN = 4
}

What kind of value a bit pattern holds.

Source

Cases

Name	Value	Description
`Zero`	`0`	Zero of either sign.
`Subnormal`	`1`	A value below the normal range, with no leading bit.
`Normal`	`2`	An ordinary finite value.
`Infinity`	`3`	Infinity of either sign.
`NaN`	`4`	Not a number, including the unnormal and pseudo encodings of the extended format.

`UnpackedWide`

pub struct UnpackedWide {
    pub class: WideClass;
    // private fields
}

A finite value taken apart: significand times two to the exponent.

Source

Fields

Name	Type	Description
`class`	`WideClass`	Which kind of value this came from.

`UnpackWide`

pub func UnpackWide(
    bits: uint512,
    layout: &FloatLayout
) -> UnpackedWide

Takes a bit pattern apart under layout.

Source

Parameters

Name	Type	Description
`bits`	`uint512`	the value's bit pattern
`layout`	`&FloatLayout`	which wide floating-point format the bits are in

Returns: the sign, significand and exponent those bits hold

`PackWide`

pub func PackWide(
    negative: bool,
    significand: uint512,
    exponent: int,
    layout: &FloatLayout
) -> uint512

Puts a finite value back together under layout, which must already be within range and precision.

Source

Parameters

Name	Type	Description
`negative`	`bool`	whether the value is negative
`significand`	`uint512`	the significand
`exponent`	`int`	the binary exponent
`layout`	`&FloatLayout`	which wide floating-point format the bits are in

Returns: the bit pattern for that sign, significand and exponent

`InfinityBits`

pub func InfinityBits(
    negative: bool,
    layout: &FloatLayout
) -> uint512

The bits of infinity under layout.

Source

Parameters

Name	Type	Description
`negative`	`bool`	whether the value is negative
`layout`	`&FloatLayout`	which wide floating-point format the bits are in

Returns: the bit pattern for an infinity of that sign

`NanBits`

pub func NanBits(layout: &FloatLayout) -> uint512

The bits of a quiet NaN under layout.

Source

Parameters

Name	Type	Description
`layout`	`&FloatLayout`	which wide floating-point format the bits are in

Returns: the bit pattern for a quiet NaN

`WriteWideFloat`

pub func WriteWideFloat(
    allocator: Allocator,
    writer: &var TextWriter,
    spec: FormatSpec,
    bits: uint512,
    layout: &FloatLayout
) -> ! FormatError

Writes bits, read under layout, as text under spec.

Source

Parameters

Name	Type	Description
`allocator`	`Allocator`	where the working storage comes from
`writer`	`&var TextWriter`	the destination
`spec`	`FormatSpec`	what the placeholder asked for
`bits`	`uint512`	the value's bit pattern
`layout`	`&FloatLayout`	which wide floating-point format the bits are in

Returns: success when the whole rendering was accepted

FixedNat

An unsigned integer of fixed capacity, holding no storage of its own.

Float128

An IEEE binary128 value, held as its bits.

Edit this page on GitHub