Struct RequestRecord 

pub struct RequestRecord {
    pub success: bool,
    pub ttft_ms: f64,
    pub e2e_ms: f64,
    pub input_tokens: u32,
    pub output_tokens: u32,
    pub output_token_count_source: OutputTokenCountSource,
    pub quality_issues: QualityIssueCounts,
    pub itl_ms: Vec<f64>,
}

One request’s measurements (input to compute_metrics).

Fields

success: bool

ttft_ms: f64

e2e_ms: f64

input_tokens: u32

output_tokens: u32

output_token_count_source: OutputTokenCountSource

quality_issues: QualityIssueCounts

itl_ms: Vec<f64>

Per-token inter-arrival times within this request (decode steps, len = output_tokens - 1). Empty if not measured.

Implementations

impl RequestRecord

pub fn tpot_ms(&self) -> Option<f64>

Per-request TPOT in ms, or None if output_tokens < 2.

pub fn meets_slo(&self, slo: &Slo) -> bool

True if all three SLO thresholds are met (TPOT is treated as met when undefined — single-token responses don’t have meaningful TPOT).

Trait Implementations

impl Clone for RequestRecord

fn clone(&self) -> RequestRecord

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Debug for RequestRecord

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

Formats the value using the given formatter.