Struct MultiReport 

pub struct MultiReport {
    pub dataset_id: Option<String>,
    pub store_kind: Option<String>,
    pub judge_fingerprint: Option<String>,
    pub metrics: Vec<MetricReport>,
    pub freshness: Option<FreshnessReport>,
}

A bundle of MetricReports with optional run metadata, suitable for JSON persistence and baseline comparison.

Fields

dataset_id: Option<String>

Free-form dataset identifier (e.g. "beir/nq" or "internal/v3").

store_kind: Option<String>

Free-form store identifier (e.g. "memvid:livetest.mv2").

judge_fingerprint: Option<String>

Opaque fingerprint of any LLM judges used. Reports with mismatched fingerprints refuse to diff to prevent silent comparison drift. Reserved for the upcoming ragas feature; pure retrieval runs leave this empty.

metrics: Vec<MetricReport>

One report per metric, in the order metrics were declared.

freshness: Option<FreshnessReport>

Optional stale/conflict freshness rollup for the same dataset.

Implementations

impl MultiReport

pub fn new(metrics: Vec<MetricReport>) -> Self

Construct a MultiReport from a metric report vector. Other metadata is filled in via the with_* builders.

pub fn with_dataset(self, id: impl Into<String>) -> Self

Attach a dataset identifier.

pub fn with_store(self, kind: impl Into<String>) -> Self

Attach a store kind identifier.

pub fn with_judge_fingerprint(self, fp: impl Into<String>) -> Self

Attach a judge fingerprint (reserved for ragas).

pub fn with_bootstrap(self, iterations: usize, level: f64, seed: u64) -> Self

Attach a deterministic percentile-bootstrap confidence interval to every metric row. Equivalent to mapping MetricReport::with_bootstrap_ci over MultiReport::metrics. See that method for the algorithm and determinism guarantees.

pub fn with_freshness(self, freshness: FreshnessReport) -> Self

Attach a freshness rollup without modifying metric rows.

pub fn with_freshness_metrics(self, freshness: FreshnessReport) -> Self

Attach a freshness rollup and append score-like freshness metrics.

Appended metric names are freshness.stale_free_rate@k and freshness.conflict_free_rate@k. Because higher is better, these rows can be gated with RegressionGate just like recall, nDCG, or MRR.

pub fn to_json(&self) -> Result<String>

Serialize as pretty-printed JSON.

pub fn to_markdown(&self) -> String

Render a compact Markdown summary table.

pub fn diff(&self, baseline: &MultiReport) -> Result<ReportDiff>

Diff this report against a baseline. Returns a ReportDiff with per-metric Δ-mean and per-query winners/losers. Fails if the two reports were produced with different judge fingerprints (silent comparison drift).

Per-query deltas are computed by intersecting the two reports’ per_query vectors on query_id. Queries missing from either side are skipped (they cannot be compared). winners, losers, and unchanged use an absolute threshold of 1e-9 to filter floating point noise; callers needing different sensitivity should inspect MetricDelta::query_changes directly.

pub fn delta_markdown(&self, baseline: &MultiReport) -> Result<String>

Render a head-to-head Markdown delta table of this report against baseline (current, baseline, and Δ = current − baseline per metric). Convenience wrapper over self.diff(baseline)?.to_markdown(); fails on a judge-fingerprint mismatch for the same reason diff does.

Trait Implementations

impl Clone for MultiReport

fn clone(&self) -> MultiReport

Returns a duplicate of the value.

fn clone_from(&mut self, source: &Self)

Performs copy-assignment from source.

impl Debug for MultiReport

fn fmt(&self, f: &mut Formatter<'_>) -> Result

Formats the value using the given formatter.

impl Default for MultiReport

fn default() -> MultiReport

Returns the “default value” for a type.

impl<'de> Deserialize<'de> for MultiReport

fn deserialize<__D>(__deserializer: __D) -> Result<Self, __D::Error>

where __D: Deserializer<'de>,

Deserialize this value from the given Serde deserializer.

impl Serialize for MultiReport

fn serialize<__S>(&self, __serializer: __S) -> Result<__S::Ok, __S::Error>

where __S: Serializer,

Serialize this value into the given Serde serializer.