b10 1.0.0

Decimal Numbers For The Rust Programming Language

The API provides a light-weight alternative to arbitrary-precision arithmetic. The const generics for base-ten exponents fix the computation steps involved at compile time. You basically get the speed of 64-bit primitives combined with the safety from big-number implementations.

let cents = b10::BaseCount::<-2>::from(199);
assert_eq!("€ 1.99", format!("€ {cents}"));

All operation is lossless by design unless explicitly stated otherwise in the name of the respective method (e.g. log10_floor). This lossless guarantee applies to parsing and formatting too. Can't read nor write digits beyond the base resolution.

// metric prefixes
use b10::{Milli, Nano, Pico};
let mA = Milli::from(100);
let ns = Nano::from(4);

// multiply into another base
let (pC, overflow):(Pico, Pico) = mA.mul(ns);
if overflow != Pico::ZERO {
    panic!("product exceeds 2⁶⁴ − 1 pico");
}

// pretty formatting options
assert_eq!(
    "0.100 × 0.000000004 = 0.000000000400",
    format!("{mA} × {ns} = {pC}"),
);
assert_eq!(
    "100E-3 × 4E-9 = 400E-12",
    format!("{mA:E} × {ns:E} = {pC:E}"),
);
assert_eq!(
    "100 mA × 4 ns = 400 pC",
    format!("{mA:#}A × {ns:#}s = {pC:#}C"),
);

Formatting and parsing performance is slighty faster than the Display and FromStr traits from the standard library.

test fmt_tests::format_exponent_kilo       ... bench:           7.81 ns/iter (+/- 0.07)
test fmt_tests::format_exponent_u64        ... bench:           9.68 ns/iter (+/- 0.36)
test fmt_tests::format_fraction_milli      ... bench:           7.82 ns/iter (+/- 0.05)
test fmt_tests::format_fraction_u64        ... bench:          19.87 ns/iter (+/- 0.37)
test fmt_tests::format_integer_natural     ... bench:           6.72 ns/iter (+/- 0.07)
test fmt_tests::format_integer_u64         ... bench:           8.59 ns/iter (+/- 0.07)
test fmt_tests::format_sub_zero_quecto     ... bench:           7.50 ns/iter (+/- 0.03)
test fmt_tests::format_sub_zero_u64        ... bench:          22.50 ns/iter (+/- 0.17)
test text_tests::parse_exponents_as_f64    ... bench:           9.14 ns/iter (+/- 0.27)
test text_tests::parse_exponents_as_pico   ... bench:           7.81 ns/iter (+/- 0.04)
test text_tests::parse_fractions_as_f64    ... bench:           7.78 ns/iter (+/- 0.04)
test text_tests::parse_fractions_as_nano   ... bench:           5.39 ns/iter (+/- 0.09)
test text_tests::parse_integers_as_centi   ... bench:           3.75 ns/iter (+/- 0.01)
test text_tests::parse_integers_as_natural ... bench:           3.75 ns/iter (+/- 0.03)
test text_tests::parse_integers_as_u64     ... bench:           3.12 ns/iter (+/- 0.01)

Function mul and pow_const require the redundant_generics feature due to uncertainty about the generic_const_exprs feature (issue #8).

[dependencies]
b10 = { version = "1.0", features = ["redundant_generics"] }

This is free and unencumbered software released into the public domain.