Seventh chart row. BT overrides MissionStarting purely to seed the score, and
the override was missing, so MESSAGE_ENTRY(BTPlayer, MissionStarting) resolved
to the inherited engine handler (which only does the fade-in) and the grant
never happened.
FUN_004bfbe8(player):
base_MissionStarting(player);
if (app->state == 4 && (player[0x29] & 0x40) == 0)
player[0x1c8] = 0x447a0000; // = 1000.0f
Both operands decode exactly against engine headers: application state 4 is
LaunchingMission (APP.h -- same enum whose 6 is EndingMission, already used by
the console flush), and simulationFlags bit 14 is NonScoringPlayerBit
(PLAYER.h: NonScoringPlayerBit = Entity::NextBit), so `(+0x29 & 0x40) == 0` IS
IsScoringPlayer(). Camera-ship/spectator players are non-scoring and correctly
get nothing.
CELL NOTE: the binary seeds the ENGINE cell (+0x1c8), not BT's own (+0x278) --
1995 carried two accumulators, which is why the KB suspected the pod's death
cost "may never have displayed". Our port has one currentScore, so grant,
awards and death cost land together and the chart reads coherently.
Also resets the console watermark so a fresh mission REPORTS the grant rather
than a difference from last round's tally.
Benched: both players "[score] mission start: player N:1 seeded to 1000",
scores run 1001.98 -> 1908.64 with kills=1 (1000 + ~400 damage + 505 kill).
Also corrects a FOURTH copy of the dead-code claim, in btplayer.hpp's ScoreType
enum ("type 0 has NO scoring arm ... per-hit inflicted credit never existed").
Its byte-scan was right that no TABLE entry binds @004c0200 and wrong to
conclude unreachable -- the vtable Dispatch override calls it directly.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01NCJQkvq6G2JNrpVbA75tVZ
