PreciseNumber.Radix returns 2, but the type is base 10 (significand × 10^exponent)

[matt-edmondson]

What's wrong

PreciseNumber/PreciseNumber.cs:437-440:

private const int BinaryRadix = 2;
/// <inheritdoc/>
public static int Radix => BinaryRadix;

INumberBase<T>.Radix is defined as "the radix, or base, for the type". PreciseNumber stores Significand × 10^Exponent and does all its digit work (significant digits, rounding, Round(decimalDigits), the rendered form) in base 10. The BCL's other decimal type, decimal, reports Radix == 10.

PreciseNumber.Test/PreciseNumberTests.cs:906-908 (TestRadix) asserts 2, so the test pins the wrong value. It does not document a design decision.

Why it matters

Generic math code uses T.Radix to reason about representation. Examples include computing ulps, deciding how many digits a value can hold exactly, choosing a digit-by-digit algorithm, or checking whether scaling by the radix is exact. With Radix == 2, such code treats PreciseNumber as binary. For example, it would conclude that x * 10 is not an exact scale, while x * 2 is.

Suggested fix

• Return 10, and rename the constant (e.g. DecimalRadix).
• Update TestRadix to expect 10.
• This is a behaviour change to a public static, so note it in the changelog.

Acceptance criteria

• PreciseNumber.Radix == 10 == decimal.Radix.