no-floating-point-equality

Disallow exact equality comparisons between floating-point values

Floating-point arithmetic rounds intermediate results, so mathematically equivalent calculations can produce different representations.
Instead, you SHOULD compare an absolute or relative error with a tolerance chosen for the value's scale and domain.

- Category: correctness
- Level: warning
- Fix: none
- Scope: module

Reported

function same(left: float64, right: float64): boolean {
    return left === right;
}
Accepted

function same(left: float64, right: float64): boolean {
    return (left - right).abs() < 0.001;
}
Prior art

- Clippy · floatcmp

language/linter/src/rules/correctness/nofloatingpointequality.rs:6