decimal-scaled - usage guides

Docs • Benchmarks • Algorithms • Roadmap • API reference

decimal-scaled provides const-generic, base-10 fixed-point decimal types with deterministic, bit-exact arithmetic. A value is stored as an integer raw such that the logical value is raw × 10^(-SCALE), where SCALE is a compile-time const generic.

These guides cover how to use the library in depth. For the high-level pitch and the "why another numeric type" comparison, see the crate README. For the full API, see the rustdoc.

Guides

Guide	What it covers
Getting started	Installing, constructing values, basic arithmetic, formatting, parsing.
The width family	D9 / D18 / D38 / D76 / D153 / D307 - when to pick which, scale ranges, the Decimal trait.
Conversions	Integer / float conversions, cross-width widening and narrowing, to_int, the float bridge.
Rounding	RoundingMode, the _with method pairs, rescale, and the compile-time rounding-* feature flags.
Strict mode	Integer-only transcendentals (ln, exp, sqrt, trig, …) under --features strict.
The d38! macro	Compile-time decimal literals with automatic scale inference, scientific notation, explicit scale, rounding, and inline expressions.
Cargo features	Every feature flag and what it enables.
Benchmarks	Head-to-head timings against bnum, ruint, rust_decimal, and fixed; fast vs strict transcendentals; the 0.5 ULP guarantee.

A 30-second tour

use decimal_scaled::{D38s2, d38};

// Compile-time literal - scale inferred from the written digits.
let price = d38!(19.99);              // D38s2, exactly 19.99
let qty   = D38s2::from_int(3);       // 3.00

let total = price * qty;               // 59.97, exact - no binary rounding
assert_eq!(total, d38!(59.97));

// Deterministic: identical bit pattern on every platform.
assert_eq!(total.to_bits(), 5997);