Entry 298

"An event's scene -2 put the town's card up: the port skipped loadCheck"

2026-09-28enginescriptui

Filescrates/piney-game/src/session.rscrates/piney-game/src/session/tests/survey.rs

Reported in play (INF): in Mac Anu during event 13 (MG0340, Mia and Elk), the "Aqua Capital / Mac Anu" loading card came up as though Kite had logged in again. Block 2 of the event (if near_marker 31 <= 80, sound, fade 10 128, wait 10, remove -1 -1, scene -2) sets the same town up again so block 3 can play stream 13. The port put the card up on every set-up of The World (Session::world_stage).

What the game does

ccFileListLoad (INF SLUS_202.67:0x00164540) calls ccLoadDispInit at 0x00164a68 only when all three hold:

• loadCheck() (0x00165530) is nonzero: some entry of the new sceneFileList (40-byte rows, name at +4) is missing from the old list (dbgList, sceneFileListNumOld rows). It returns 1 at the first new file, 0 when every file was already there;
• the thread's +0x18 is 0. ccLoadFLStart (0x001654a0), the start for ccLoadFLAdd/ccLoadFLAddOne, stores 1 there, so extra loads never show a card;
• ccLoadDispCheck() (0x0019c430) is 0: no display is already up.

A scene -2 reload of the same town reads nothing new, so the game shows no card; the screen stays black from the script's fade until the town fades back in.

The fix

world_stage already recorded what each set-up inflates, for --dvd's timing (resident). When every file the new set-up read is in the last set-up's set, the card is taken down before it draws. Session::change (any ChangeRequest: desktop, board, title, log in) now clears resident, because those modes load their own lists and the game's old list is then theirs, not the last area's.

event_13_s_town_again_has_no_card puts Kite at marker 31 once block 1 is done and fails on any card before stream 13 plays. Without the fix it fails 21 frames after the put ("loading 0"). The piney-game suite passes (162).