Struct PhaseProgressUpdate 

pub struct PhaseProgressUpdate {

    pub exec_id: u64,
    pub name: String,
    pub labels: String,
    pub cursor_name: String,
    pub cursor_extent: u64,
    pub daemon: bool,
    pub rows_consumed: u64,
    pub rows_total: u64,
    pub fibers: usize,
    pub ops_started: u64,
    pub ops_finished: u64,
    pub ops_ok: u64,
    pub skips: u64,
    pub errors: u64,
    pub retries: u64,
    pub ops_per_sec: f64,
    pub adapter_counters: Vec<(String, u64, f64)>,
    pub rows_per_batch: f64,
    pub relevancy: Vec<RelevancyLive>,
}

Live metrics snapshot for progress updates.

Fields

exec_id: u64

The execution this update belongs to (SRD-88 exec_id, SRD-100 §4). With name+labels it forms the live routing key, so concurrent executions of the same phase route to distinct ActivePhase slots. 1 for the single-execution case (SRD-88 A1).

name: String

Phase name this update belongs to — matches the name passed to RunObserver::phase_starting. Present so observers that track multiple concurrent phases can route the update to the correct per-phase slot.

labels: String

Phase dimensional labels (e.g. profile=label_00, k=10) — together with name this uniquely identifies one phase iteration.

cursor_name: String

cursor_extent: u64

daemon: bool

SRD-82 Part 6 — whether this phase runs as a DAEMON (off the foreground budget, open-extent cursor, stopped by its group’s foreground completing). Display consumers exclude daemon phases from AGGREGATE progress — the shared footer gutter bar averages the non-daemon phases — since a daemon’s percent-of-budget is not meaningful progress toward the run.

rows_consumed: u64

Cursor ordinals CONSUMED so far for a data-driven phase (polydat DataSourceFactory::global_consumed()) — the authoritative row-level progress. 0 for non-cursor phases (plain cycles:), where the display keeps the op-denominated cycles: chip.

rows_total: u64

Cursor ordinal EXTENT for a data-driven phase (global_extent()). 0 for non-cursor phases — distinct from cursor_extent, which falls back to the phase’s cycles: bound for sourceless phases; this stays 0 so the display can tell a declared cursor from the synthesized range(0, cycles) every phase gets. rows_total > 0 is the “show rows:” signal.

fibers: usize

ops_started: u64

ops_finished: u64

ops_ok: u64

skips: u64

SKIPPED ops (skips_total) — excluded from the ok% denominator (a skip is neither a success nor a failure).

errors: u64

retries: u64

ops_per_sec: f64

adapter_counters: Vec<(String, u64, f64)>

rows_per_batch: f64

relevancy: Vec<RelevancyLive>

Live relevancy aggregates — one entry per relevancy metric (e.g. recall@10). Each has a moving-window mean over the last N recall calculations and a whole-activity running mean.

Trait Implementations

impl Clone for PhaseProgressUpdate

fn clone(&self) -> Self

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Debug for PhaseProgressUpdate

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

Formats the value using the given formatter.