Skip to content

Recent lap review — source-only QA (v0.5.70)

What is implemented

The real Unreal Results page now displays the eight most recent positively timed lap records (newest first) from the same bounded RaceProgress::GetRecentTimedLap ring already used for race insights. Each visible row contains the true physical completed-lap number and time. Results also identify an equal run best or the nonnegative seconds behind the fastest observed lap in this run. On desktop, a second line shows signed pace change compared with the preceding stored timed lap: positive means slower; negative means faster.

The menu's existing up/down detail pagination exposes rows that will not fit in the safe on-screen Android/desktop lane. It does not remove, scroll, or overwrite existing checkpoint, lap, reward, control, or save state.

Hard correctness constraints

  • All results are derived from the existing completed race. No synthetic opponent times, sector extrapolations, or saved all-time ghost records.
  • If CanAwardFinishedRace fails (incomplete, missing active lap timing, non-recordable clocks), the review is hidden in its entirety.
  • A zero-time or penalty-only lap cannot appear as a legitimate result.
  • Each row's lap number must be positive, ordered newest-to-oldest, and no larger than the completed-lap count; corrupt data fails closed.
  • Every accepted duration must pass IsValidFinishedRaceTime; negative best-lap gaps fail closed, rather than implying a faster result than the source-of-truth best.
  • The oldest surviving record never shows an invented "previous" lap delta when its predecessor has fallen off the eight-lap ring.
  • The review uses a fixed eight-element std::array. It does not allocate dynamic game-state storage or touch save schema, achievements, career credits, race clock, penalties, car physics, or rival policy.

Source validation

On a portable C++20 toolchain:

cmake -S Tests/Standalone -B Tests/Standalone/build -DCMAKE_BUILD_TYPE=Release
cmake --build Tests/Standalone/build --config Release
ctest --test-dir Tests/Standalone/build -C Release -R '^(RecentLapReview|MenuDetailPager|FinishedRaceInsights)$' --output-on-failure
python3 -m unittest discover -s Tests/Editor -p '*LapReviewBindingTests.py' -v

The RecentLapReview test compiles the real production RaceProgress.cpp and verifies a 10s/8s/12s three-lap finish, best-lap gaps, signed change, a single-lap finish, incomplete results, penalty-only finishes, invalid zero-clock laps, reset and a twelve-lap event with eight retained history records. Existing paged-menu tests keep scrolling and narrow-viewport text behavior separate from racing rules.

Unreal and hardware qualification still required

Portable compilation and source binding tests do not prove that Unreal Engine 5.8 compiles or renders the HUD. On an authorized UE 5.8 installation, compile LatebrakeEditor/UHT and game targets, then:

  1. Cook the authentic original circuit and package actual Win64/Android candidates. Record the exact binary/map/source fingerprints.
  2. Complete a three-lap race on Windows. Open Results; verify real lap numbers, ascending/descending signed deltas and "RUN BEST".
  3. Check a one-lap event and a longer event; the history must remain bounded to eight and never claim unavailable earlier timings.
  4. Resize windows and rotate physical Android devices. Verify UP/DOWN pads access all details; clipped strings, touch pads, SAVE FAILED warnings, title and status remain legible and operable.
  5. Try interrupted app focus/menu transitions and restart. Confirm results never leak across runs and no race authority changes on page navigation.
  6. Log hardware, Unreal revision, test date and actual failures in Android qualification issue #13.

Do not ship as a playable release or attach fabricated executables. The latest verified public pre-release remains source-only v0.5.65 until a separate tag qualification successfully publishes a newer source snapshot.