f64

64-bit floating-point numbers, f64: binary64’s own constants, its safe integers, conversion from the other shapes, and conversion from text, the same in every locale. The functions that every float format shares, such as float::sqrt and float::format, are in float, and abs, sign, and range are the prelude’s.

Write these names after f64::, as in f64::min_safe_int.

min_safe_int

def min_safe_int! f64

Smallest integer whose neighboring integers remain exactly representable.

max_safe_int

def max_safe_int! f64

Largest integer whose neighboring integers remain exactly representable.

epsilon

def epsilon

The gap between one and the next larger number; not a universal comparison tolerance.

finite

set finite = f64:{-1.7976931348623157e308..1.7976931348623157e308}

Every number but the infinities and nan.

pi

def pi

The binary64 approximation of the ratio of a circle’s circumference to its diameter.

e

def e

The binary64 approximation of the base of natural logarithms.

from

def from[set T: numeric](value! T)! f64

The f64 nearest a number of any shape, rounding once, ties to even: an int from its exact Value, so an int beyond 2^53 may round, and an f32 exactly. An f64 stays as it is.

is_safe_int

def is_safe_int(value! f64)! bool

True for finite integers in the inclusive min_safe_int–max_safe_int range.

parse

def parse(text! str)? f64

Parses a complete signed decimal/scientific spelling, or exactly nan, inf, or -inf, into one f64, and gives an empty Bag for any other text: f64::parse(".5") is 0.5, f64::parse("nan") is nan, and f64::parse("x") is empty. Accepts +1, .5, 1., and exponents; at least one mantissa digit is required. Whitespace, separators, radix prefixes, and trailing text are not numbers. Conversion rounds once to binary64, allows overflow to infinity, and canonicalizes negative zero. It takes one text, so a Bag of texts is parsed with map.