Skip to content

Unofficial quick-race rival finish-clock comparison — source QA

Status: Unreal source integrated; NOT engine/runtime/hardware-qualified. The existing Unreal Engine 5.8 single-player Quick Race Circuit contains one autonomous rival with its own physically verified ordered checkpoint trace. Previously, Results could only observe whether the rival's route was already complete at the player's verified finish.

This source update additionally captures the rival's accepted final checkpoint time on the same local active post-GO clock. After a recordable player finish, it can show an approximate observation:

  • UNOFFICIAL: RIVAL FINAL GATE OBSERVED X.XXs EARLIER when the stored rival final gate was over 0.25 seconds earlier on the sampled active clock.
  • UNOFFICIAL: FINAL GATES TOO CLOSE TO TIME for a nonnegative difference of 0.25s or less. This avoids interpreting the order of two engine tick callbacks as a certified side-by-side finish.
  • No margin when the rival has not finished, the player finish timing is invalid, the clock is malformed, or the comparison is ineligible.

No declared winner/rank/placement. Unreal actor tick order, discrete physics samples and local clock granularity make the observation insufficient to declare an official winner. The current career race, Time Attack, credits, championship medals, best laps, official clock, player gate acceptance and SaveGame do not depend on this value. The separate rival-route-completion message remains available.

Integrity and lifecycle

  • RivalFinishClockReview.h is a pure allocation-free C++20 policy, configured using the actual race checkpoint count and required laps.
  • The rival observer runs only after its own forward directional gate trace and matching split policy have accepted the gate; recording additionally requires its route-complete state.
  • Rival and player timestamps remove player recovery penalty fees from the race elapsed value to compare a common active physical session clock, not a penalty-adjusted official race outcome.
  • Player completion is checked with ValidateFinishTiming() before capturing Results. The comparison requires Quick Race Circuit, a separately verified physically ready rival, exact matching gate count and legitimate final player ordinal.
  • Race initialization/restarts reset the private rival timestamp and Results snapshot. Duplicates, backdated clocks, wrong event sizes, nonfinite values, out-of-range time and stale callbacks cannot overwrite a prior accepted final gate.

Source-level verification

The standalone RivalFinishClockReviewTests.cpp exercises valid observed margins, close-call uncertainty, invalid/before-final player states, repeated rival callbacks, corrupted clocks, wrong gate counts, seven-day upper bound and reset isolation on strict portable C++20.

UnrealRivalFinishClockBindingTests.py asserts that GameMode and Results actually consume this policy in the intended order and cannot change award or save authority.

Both are source-only tests, not physical race tests.

Genuine engine acceptance still required

  1. Compile with real licensed Unreal Engine 5.8/UHT and package Win64.
  2. Drive Quick Race Circuit against a physically moving single rival. Observe the rival cross its final directional gate first and compare the displayed margin with captured session and race audit logs.
  3. Test a close final-gate crossing: the UI must present uncertainty.
  4. Confirm solo Time Attack has no rival comparison and careers retain their previous independent scoring.
  5. Restart, pause/reconnect, recover, and change race/event: old finish observations must not persist.
  6. Confirm results layout, pagination and localization on installed Windows/Android builds before shipping.

See playable-release gate #148.