Trait float_eq::FloatEqUlpsTol
source · [−]Expand description
Per-field tolerances for ULPs comparisons.
This trait establishes a one-to-one relation between an IEEE floating point type and a type whose fields are expected to be structurally identical but specified in ULPs. It is used by ULPs equality checks to specify per-field tolerances.
By convention, this type is named FooUlps
for a given type Foo
.
The UlpsTol
type alias exists to simplify usage, for example UlpsTol<f32>
is u32
.
To implement this trait over a new type, see How to compare custom types.