Struct FiniteScoreSet 

pub struct FiniteScoreSet<T: Float, I, E> { /* private fields */ }

A weighted set of scoring operators using enum-based static dispatch.

FiniteScoreSet stores a Vec of FiniteMembers, each wrapping a variant of a user-declared metric enum. At evaluation time, the enum’s eval method dispatches via match — zero vtable overhead for all non-Custom variants.

Construct via finite_score_set!, the FiniteScoreSetBuilder, or call .score() directly.

Type parameters

• T: Float — the floating-point type (f32 or f64).
• I — the input type passed to each metric.
• E: Scorable<T, I> — the metric enum generated by finite_metric!.

Example

let set = FiniteScoreSet::<f64, &str, TestKind<f64, &str>>::normalize(vec![
    (2.0, TestKind::AlwaysZero(ConstMetric::new("zero", 0.0))),
    (3.0, TestKind::AlwaysOne(ConstMetric::new("one", 1.0))),
])?;

let total = set.sum(&"input");
// total = 0.4 * 0 + 0.6 * 1 = 0.6

Implementations

impl<T: Float, I, E: Scorable<T, I>> FiniteScoreSet<T, I, E>

pub fn sum(&self, input: &I) -> T

Evaluate all metrics against input and sum their weighted contributions.

This is the most common aggregation: each metric is evaluated, multiplied by its normalized weight, and summed. Zero-allocation convenience.

For custom aggregation, use .score() instead.

pub fn score(&self) -> FiniteScoreStage<'_, T, I, E>

Enter the scoring stage, returning a reference to all members.

Use .by() on the returned stage to apply a custom aggregation, or .sum() for the standard weighted-sum shortcut.

Example

let total = set.score().by(|members| {
    members.iter().fold(0.0, |acc, m| {
        acc + m.contribute(m.metric().eval(&input))
    })
});

pub fn len(&self) -> usize

Return the number of members in this set.

pub fn is_empty(&self) -> bool

Return true if the set has no members.

pub fn iter(&self) -> impl Iterator<Item = &FiniteMember<T, E>>

Iterate over the members.

pub fn breakdown(&self, input: &I) -> Vec<Breakdown<'_, T>>

Evaluate all metrics against input and return a per-metric breakdown.

Unlike .sum() which returns only the aggregate, breakdown returns one Breakdown row per member with the metric’s name, raw score, normalized weight, and weighted contribution.

pub fn builder() -> FiniteScoreSetBuilder<T, I, E>

Create a builder for incremental construction of a FiniteScoreSet.

Use this when members are not known up front — push them one by one, then call .build() to finalize.