Struct PerRequestRecord 

pub struct PerRequestRecord {

    pub session_id: Option<String>,
    pub turn_index: Option<usize>,
    pub uuid: String,
    pub arrival_time_ms: f64,
    pub first_admit_ms: Option<f64>,
    pub terminal_time_ms: f64,
    pub first_token_ms: Option<f64>,
    pub last_token_ms: Option<f64>,
    pub ttft_ms: Option<f64>,
    pub ttst_ms: Option<f64>,
    pub e2e_latency_ms: Option<f64>,
    pub itl_ms: Option<f64>,
    pub input_length: usize,
    pub requested_output_length: usize,
    pub output_length: usize,
    pub reused_input_tokens: usize,
    pub prefill_worker_idx: Option<usize>,
    pub decode_worker_idx: Option<usize>,
    pub prefill_admit_ms: Option<f64>,
    pub source_held_ms: Option<f64>,
    pub destination_reserved_ms: Option<f64>,
    pub destination_activated_ms: Option<f64>,
    pub decode_admit_ms: Option<f64>,
    pub source_released_ms: Option<f64>,
    pub decode_reused_input_tokens: Option<usize>,
    pub prefill_route_overlap_tokens: Option<usize>,
    pub decode_route_overlap_tokens: Option<usize>,
    pub terminal_status: ReplayTerminalStatus,
}

Flat per-request record for --report-jsonl emission. One JSON line per request in the JSONL output; consumed by external analysis tools that want per-request granularity (TTFT vs. ISL scatter, worker-residency analysis, bypass classification, etc.).

Fields

session_id: Option<String>

Session identifier from the trace, when present. Mirrors AIPerf’s conversation_id field for the same purpose: bucket per-request records by multi-turn session. Placed first in the serialized output so each JSONL row leads with its session/turn identity, matching AIPerf’s profile_export.jsonl layout.

turn_index: Option<usize>

Zero-based turn index within session_id, when present.

uuid: String

arrival_time_ms: f64

first_admit_ms: Option<f64>

terminal_time_ms: f64

first_token_ms: Option<f64>

last_token_ms: Option<f64>

ttft_ms: Option<f64>

ttst_ms: Option<f64>

e2e_latency_ms: Option<f64>

itl_ms: Option<f64>

Inter-token latency for this request, in milliseconds. Matches AIPerf’s inter_token_latency field — one scalar per request.

input_length: usize

requested_output_length: usize

Number of output tokens requested by the workload trace.

output_length: usize

Number of output tokens actually emitted by the mock engine.

reused_input_tokens: usize

prefill_worker_idx: Option<usize>

decode_worker_idx: Option<usize>

prefill_admit_ms: Option<f64>

source_held_ms: Option<f64>

destination_reserved_ms: Option<f64>

destination_activated_ms: Option<f64>

decode_admit_ms: Option<f64>

source_released_ms: Option<f64>

decode_reused_input_tokens: Option<usize>

prefill_route_overlap_tokens: Option<usize>

decode_route_overlap_tokens: Option<usize>

terminal_status: ReplayTerminalStatus

Trait Implementations

impl Clone for PerRequestRecord

fn clone(&self) -> PerRequestRecord

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Debug for PerRequestRecord

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

Formats the value using the given formatter.

impl Serialize for PerRequestRecord

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

where __S: Serializer,

Serialize this value into the given Serde serializer.