LS1008: Domain Uses Binary Floating Point

LS1008 is a compiler check that reports a double or float type in Domain, the project that holds business rules, because binary rounding makes totals and equality checks on money and exact quantities quietly wrong. This page states what the rule reports, where the warning sits, and how to hold the value in decimal or a value object.

Reference#

What the rule enforces#

LS1008 enforces Keep binary floating point out of exact quantities (litespec/backend/domain#keep-binary-floating-point-out-of-exact-quantities). It runs in Domain only. Its category is LiteSpec.Domain. It reports a warning and is enabled by default, as the Analyzers overview describes.

What the rule reports#

The rule reports each double or float type in a field, property, indexer, method return, local function return, delegate, operator, parameter, foreach variable, declaration expression, or declaration pattern, including nested uses such as List<double>, double?, tuples, and lambda parameters. The diagnostic sits on the floating point type. Its message names the member and the type text.

Failing example#

public sealed class Pricer
{
    public double Total(double rate) => rate;
}

Passing example#

public sealed class Pricer
{
    private decimal _total;

    public decimal Total(decimal rate) => rate;
}

How to fix#

Use decimal or a value object for monetary amounts, rates applied to them, and quantities compared for exact equality.

Notes#

What the rule ignores#

The rule does not report decimal or var. It does not report any floating point use outside Domain.