Race feed (LAN): names drop after finish, bogus best lap, inconsistent race_result

mintopia commented

2026-10-20 01:05:24 +00:00

Owner

Found while verifying #27 on a 5-instance LAN race (Barcelona Oval, 3 laps, BLUR_LANSIM, recomp). Captures are local (ws_tap.py record, not in git).

Names vanish after finish: once the local player finishes, every racer's name/car/mods drop out of standings (all 5 racers, ~70 s on the winner's instance), and race_result standings never carry them. The live-timing classification then shows CPU 0x / Player.
Bogus best lap: one remote racer had best_ms 100 (page showed fastest lap 0:00.100 L1).
race_result inconsistent across instances: only 2 of 5 instances sent "final":true; on one instance a finished racer in P2 had a lower race_ms (154723) than the P1 winner (188122).
Transient unknown powerup: one remote racer's slot read as an out-of-range kind for ~1 s on one instance (once per 5-instance race).

Acceptance criteria

Names/cars/mods present in standings and race_result after finishing
best_ms never below a plausible lap time
race_result final and consistent on every instance

Found while verifying #27 on a 5-instance LAN race (Barcelona Oval, 3 laps, `BLUR_LANSIM`, recomp). Captures are local (`ws_tap.py record`, not in git). - **Names vanish after finish**: once the local player finishes, every racer's `name`/`car`/`mods` drop out of `standings` (all 5 racers, ~70 s on the winner's instance), and `race_result` standings never carry them. The live-timing classification then shows `CPU 0x` / `Player`. - **Bogus best lap**: one remote racer had `best_ms` 100 (page showed fastest lap `0:00.100 L1`). - **`race_result` inconsistent across instances**: only 2 of 5 instances sent `"final":true`; on one instance a finished racer in P2 had a lower `race_ms` (154723) than the P1 winner (188122). - **Transient `unknown` powerup**: one remote racer's slot read as an out-of-range kind for ~1 s on one instance (once per 5-instance race). ## Acceptance criteria - [ ] Names/cars/mods present in `standings` and `race_result` after finishing - [ ] `best_ms` never below a plausible lap time - [ ] `race_result` final and consistent on every instance

mintopia added the

needs-triage

label

2026-10-20 01:05:24 +00:00

mintopia referenced this issue

2026-10-20 01:05:36 +00:00

Live timing: verify POWERUPS column against a real capture #27

mintopia self-assigned this

2026-10-20 04:10:06 +00:00

mintopia commented

2026-10-20 04:10:55 +00:00

Author

Owner

Fixed in race_feed: (1) set_players merges by racer, so names/cars/mods survive the game dropping remote records after finish (cleared at teardown/reset); (2) laps/best under 5 s ignored (kMinLapS) so bogus best_ms 100 is rejected; (3) teardown now publishes final:true if a finish was seen but final was not, and finished racers race_ms is clamped monotonic by finish order. Unit test extended. Not addressed: transient unknown powerup (not in acceptance criteria). Not verified on a live LAN capture.

mintopia commented

2026-10-20 04:12:08 +00:00

Author

Owner

Follow-up: finished racers race_ms now = race clock at first observed finish (not cum of lap hops, which miss laps); local race_ms_ frozen from same; reset() publishes final:true if finished but not final. Unit tests cover it. Live 5-instance LAN capture not run (no game/LAN in this environment).

mintopia closed this issue

2026-10-20 04:12:51 +00:00

mintopia commented

2026-10-20 04:12:51 +00:00

Author

Owner

Completed and merged by Harmonic (task 28).

mintopia referenced this issue from a commit

2026-10-20 08:39:01 +00:00

Race feed: keep player names after finish, reject implausible laps, always send final race_result, order finished race_ms (#28)

mintopia referenced this issue from a commit

2026-10-20 08:39:01 +00:00