trueskill-tt/tests/ingestion_equivalence.rs

//! Ingesting the same events must give the same answer however they were
//! batched.
//!
//! The numerical goldens all ingest in a single call with one slice per
//! timestamp, so they never exercise the "append to an existing slice" path.
//! These do.

use smallvec::smallvec;
use trueskill_tt::{ConvergenceOptions, Event, Gaussian, History, Member, Outcome, Team};

/// Converge tightly: the default cap of 30 iterations leaves a residual around
/// 1e-6, which would swamp the comparison. Both paths must reach the same
/// fixed point, so drive both well past it.
fn tight() -> ConvergenceOptions {
    ConvergenceOptions {
        max_iter: 2_000,
        epsilon: 1e-12,
        ..ConvergenceOptions::default()
    }
}

fn event(a: &str, b: &str, time: i64) -> Event<i64, String> {
    Event {
        time,
        teams: smallvec![
            Team::with_members([Member::new(a.to_string())]),
            Team::with_members([Member::new(b.to_string())]),
        ],
        outcome: Outcome::winner(0, 2),
    }
}

/// Like [`event`], but `a` carries competitor configuration.
///
/// `prior` and `drift_scale` configure the competitor rather than the event, so
/// they are the part of ingestion most exposed to order: they are consumed once,
/// where the competitor's state is written.
fn configured_event(a: &str, b: &str, time: i64, scale: f64) -> Event<i64, String> {
    Event {
        time,
        teams: smallvec![
            Team::with_members([Member::new(a.to_string()).with_drift_scale(scale)]),
            Team::with_members([Member::new(b.to_string())]),
        ],
        outcome: Outcome::winner(0, 2),
    }
}

fn converged_skills(events: Vec<Event<i64, String>>, batched: bool) -> Vec<(String, Gaussian)> {
    let mut h: History<i64, _, _, String> = History::builder()
        .key_type::<String>()
        .convergence(tight())
        .build();

    if batched {
        h.add_events(events).unwrap();
    } else {
        for ev in events {
            h.add_events(std::iter::once(ev)).unwrap();
        }
    }

    let report = h.converge().unwrap();
    assert!(
        report.converged,
        "fixture must converge before results can be compared; final step {:?}",
        report.final_step
    );

    let mut skills: Vec<(String, Gaussian)> = h
        .learning_curves()
        .into_iter()
        .map(|(key, curve)| (key, curve.last().unwrap().1))
        .collect();
    skills.sort_by(|a, b| a.0.cmp(&b.0));
    skills
}

fn assert_same(batched: &[(String, Gaussian)], incremental: &[(String, Gaussian)], what: &str) {
    assert_eq!(
        batched.len(),
        incremental.len(),
        "{what}: competitor count differs"
    );

    for ((kb, gb), (ki, gi)) in batched.iter().zip(incremental.iter()) {
        assert_eq!(kb, ki, "{what}: key order differs");
        assert!(
            (gb.mu() - gi.mu()).abs() < 1e-8 && (gb.sigma() - gi.sigma()).abs() < 1e-8,
            "{what}: {kb} differs — batched mu={} sigma={}, incremental mu={} sigma={}",
            gb.mu(),
            gb.sigma(),
            gi.mu(),
            gi.sigma()
        );
    }
}

/// All events share one timestamp, so incremental ingestion repeatedly appends
/// to an existing slice.
#[test]
fn same_slice_incremental_matches_batched() {
    let events = vec![
        event("a", "b", 1),
        event("c", "d", 1),
        event("e", "f", 1),
        event("a", "c", 1),
        event("b", "e", 1),
    ];

    let batched = converged_skills(events.clone(), true);
    let incremental = converged_skills(events, false);
    assert_same(&batched, &incremental, "single shared slice");
}

/// Distinct timestamps, so each append lands in a fresh slice appended after
/// the existing ones.
#[test]
fn distinct_slices_incremental_matches_batched() {
    let events = vec![
        event("a", "b", 1),
        event("b", "c", 2),
        event("c", "a", 3),
        event("a", "c", 4),
    ];

    let batched = converged_skills(events.clone(), true);
    let incremental = converged_skills(events, false);
    assert_same(&batched, &incremental, "distinct slices");
}

/// Several events per timestamp across several timestamps — appends to
/// existing slices interleaved with new ones.
#[test]
fn mixed_slices_incremental_matches_batched() {
    let events = vec![
        event("a", "b", 1),
        event("c", "d", 1),
        event("a", "c", 2),
        event("b", "d", 2),
        event("a", "d", 3),
        event("b", "c", 3),
    ];

    let batched = converged_skills(events.clone(), true);
    let incremental = converged_skills(events, false);
    assert_same(&batched, &incremental, "mixed slices");
}

/// Appending an event to a slice that is *not* the most recent one exercises
/// the forward refresh of every later slice.
#[test]
fn back_dated_event_matches_batched() {
    let events = vec![
        event("a", "b", 1),
        event("b", "c", 5),
        event("c", "a", 9),
        // arrives last, but belongs to the middle slice
        event("a", "c", 5),
    ];

    let batched = converged_skills(events.clone(), true);
    let incremental = converged_skills(events, false);
    assert_same(&batched, &incremental, "back-dated event");
}

/// The invariant this file protects was only ever checked for *unconfigured*
/// competitors — every helper above built members with `Member::new`.
///
/// Configuration is the part most exposed to ordering, because it is consumed
/// once at the point the competitor's state is written rather than replayed per
/// event. These cover it.
#[test]
fn configured_competitors_are_order_independent() {
    let events = vec![
        configured_event("a", "b", 0, 0.0),
        configured_event("a", "c", 1, 0.0),
        configured_event("a", "b", 2, 0.0),
        event("b", "c", 3),
    ];

    assert_same(
        &converged_skills(events.clone(), true),
        &converged_skills(events, false),
        "configuration repeated on every appearance",
    );
}

/// Configuration supplied only on a *later* event is the case that used to be
/// silently dropped. It must now reach the same fit either way it is ingested.
#[test]
fn late_configuration_is_order_independent() {
    let events = vec![
        event("a", "b", 0),
        configured_event("a", "c", 1, 0.0),
        event("a", "b", 2),
    ];

    assert_same(
        &converged_skills(events.clone(), true),
        &converged_skills(events, false),
        "configuration supplied after first appearance",
    );
}

/// And it must actually be doing something — an implementation that dropped
/// configuration entirely would pass both tests above.
#[test]
fn configuration_changes_the_fit_however_it_is_ingested() {
    let configured = vec![
        event("a", "b", 0),
        configured_event("a", "c", 1, 0.0),
        event("a", "b", 2),
    ];
    let plain = vec![event("a", "b", 0), event("a", "c", 1), event("a", "b", 2)];

    for batched in [true, false] {
        let with = converged_skills(configured.clone(), batched);
        let without = converged_skills(plain.clone(), batched);
        assert!(
            with.iter()
                .zip(&without)
                .any(|((_, x), (_, y))| (x.sigma() - y.sigma()).abs() > 1e-9),
            "batched={batched}: configuration had no effect, so the order tests are vacuous"
        );
    }
}