Files
Viano-Core/doc/clientless/checkpoints/2026-10-29-deck-navigation/validation.md
T

3.6 KiB

Deck navigation validation

Build and installation succeeded with the existing Debug configuration. Added private Recast dependency; no external data extraction/reinstallation needed. Fresh worldserver PID270232 reached ready and listens on TCP8085. Console: /tmp/vcore-deck-start-20261029.log. Six targeted test programs passed.

Native diagnostics:

  • Bravery176310:1678 polygons,259 reachable sample targets,259 return paths.
  • Subway176081:105 polygons,59 reachable sample targets,59 return paths.

Controlled live ship journey, Renaisia70:

  • 11:36:57 Boarding at ship_sw_br.
  • 11:37:04 Riding/passenger1; DECK_WALK2m,3 points.
  • Native saveall during transit persisted local position (-1.74993,7.03908,6.14418), matching the mesh walk target.
  • Native inspect simultaneously reported stopped0 on The Bravery.
  • 11:38:42 DECK_WALK0.79m back to the recorded exit anchor.
  • 11:38:42 Leaving;11:38:44 Arrived/passenger0 at ship_ruttheran_br.
  • Temporary manual test schedule override restored to ON and acknowledged.

Noralin80 was positioned offline using native tele name at DeeprunTram and logged in for a second live check. Her normal schedule takes tram_if to tram_sw; no manual schedule override. Native journey completed: 11:40:44 Boarding,11:40:47 Riding/passenger1;11:41:55 mesh return1.07m to Stormwind exit anchor, followed by11:41:57 Arrived/passenger0. Initial spreading was skipped because of an overly conservative departure budget; adjusted the budget to outward duration+2s, reserving return/exit time at the destination. The updated production scheduling regression passes.

No visual acceptance claimed; native geometry, controller tests and logs are separate evidence. Remaining user acceptance: multiple visible passengers, clothing/body sizes, interruption close to departure and crowded boarding.

Final build3 was installed after another verified clean shutdown. Fresh console /tmp/vcore-deck-final-20261029.log reached ready; PID270857, TCP8085 listening. The diagnostic for subway entry176080 also confirms105 polygons/59 return paths, sharing the same model cache as176081.

Final live subway check, Noralin80 (normal schedule, no manual override):

  • Native offline tele name to DeeprunTram, followed by clientless login80.
  • 11:48:00 Boarding,11:48:02 Riding/passenger1 and outward DECK_WALK2.00m.
  • Native saveall persisted local(-2.1012,-0.396121,-10.3489), matching the target. Native inspect reported stopped0 and distance10.57 from the wagon origin, consistent with that local offset while the vehicle was travelling.
  • 11:49:13 return mesh walk2.02m;11:49:19 Leaving.
  • 11:49:26 vehicle departed during the recorded exit; existing safety retained the passenger and entered Riding/abort pending a safe recorded dock.
  • This final journey does NOT count as a successful full exit. Prior ship and subway journeys completed, but the final stress run exposed exit timing under heavy world-update stalls (CPC_PERF reported50830ms with8 ticks). Normal travel duration reserves cannot guarantee completing a multi-step exit across such stalls. No vehicle schedule was altered to hide this limitation.

Final source diff check passed. No schema/config changes, no commit/push. Ship70's manual schedule override is restored; Noralin80 remains under her normal schedule and the existing safe transport cancellation/recovery logic.

Recovery completed at11:51:36: native Leaving followed by Failed/passenger0 with 'cancelled at safe dock'. The bot was safely detached on land and normal day-plan travel resumed. This is successful safety recovery, not an uninterrupted journey. No manual override remains.