umath

C++23 numeric building blocks:

• uint128 / int128: portable 128-bit integers (uses __int128 when available, otherwise pure C++).
• Numeric128: fixed-scale decimal backed by int128, with precision P (<= 38) and compile-time scale S.
• Numeric: arbitrary-precision decimal using base-1e9 limbs, with PostgreSQL-like limits:
• up to 131072 digits before the decimal point
• up to 16383 digits after the decimal point

Design

• No exceptions in core types (errors are stored as Err).
• Rounding modes: Trunc, HalfUp.
• Deterministic to_string() / parsing.

Quick example

#include <iostream>
#include "umath/NumericFixed.h"

using usub::umath::Numeric;
using usub::umath::Numeric128;
using usub::umath::Rounding;

int main() {
  Numeric128<38,4> a("12.34005", Rounding::HalfUp);
  Numeric128<38,4> b("2.5000");
  std::cout << (a * b).to_string() << "\n"; // 30.8501

  Numeric x("1.23");
  Numeric y("2.5");
  std::cout << (x + y).to_string() << "\n"; // 3.73
}