Entry 111

The scripts' last host gaps: a companion's item_add, virus_core's level, room by RoomSelect, save_party

2026-09-25scriptworldtest

Filescrates/piney-game/src/field_host.rscrates/piney-game/src/area_host.rscrates/piney-game/src/area.rscrates/piney-game/src/world.rscrates/piney-game/src/session.rscrates/piney-fieldui/src/lib.rscrates/piney-world/src/field_world.rscrates/piney-world/src/dungeon_area.rscrates/piney-world/examples/dungeon_probe.rstools/test_dungeon_rt.pydocs/engine/event-vm.mddocs/engine/dungeon.md

A static survey of Infection's scripts found four instructions whose Host methods still fell to the defaults: item_add (add_spc_item), virus_core (generate_area), room and room_point (room), and save_party in M130's block 20. I checked each against the game's own case in ccEvent::Execute first. Two of them turned out never to reach the host where the survey placed them.

item_add

Case 93 (main 0x001afd04):

• At level 1, for Kite (pc 0), or for a category of 10 and up, it calls ccSaveData::AddItem.
• Otherwise it calls ccEvent::GetSpc(pc). If that finds a character, ccMenu->AddSpcItem(ch, cat, id, num, 0) runs instead, and the item does not go into the save.

M119's item_add 0 15 14 1 is for Kite, so it never asks the host. Neither does any other volume-1 item_add to Kite.

The companions' items are in dungeons. S108 gives Sanjuro (4) the Kotetsu Sword (2/80), and S111 gives Natsume (11) item 0/60. Both are players met in their own dungeons.

The port.

• FieldUi::add_spc_item runs the ported AddSpcItem (menus/talk.rs) with tf 0, so it never breathes. It wears the item if it is better, reads it if it is a book, and otherwise puts it in the bag.
• The town's host looks up the member in TownParty. The area's host looks up the battle character (Combat::who). If there is none, the host declines and the interpreter puts the item in the save.

virus_core

Case 124 (0x001b06c8) returns at once unless the level is exactly 1: slti $at, $s5, 2; beqz $at skips level 2 and above. At level 1 it does three things:

• SimGenerateCode from the area's three word IDs;
• sets the protectArea bit;
• deletes the area's protect items (category 15) from Kite.

Level 1 means ccEventFlagSet. That is a volume's events brought forward, or the port's story starts, and start.rs's replay host already answers those. So M128 in Mac Anu never asks the town's host while playing.

The hosts implement it anyway, as WORLD_MAN would be left: the area the words make on the current server, via FieldUi::set_area_words.

room and room_point

Cases 130 and 131 play only. They call WORLD_MAN::RoomSelect(floor, block) (main 0x0019dca0). room_point uses the room of the first event point with that number (already in the VM).

RoomSelect runs, in order:

• bgColor 0 and the two GS words;
• ClearRoom and SetRoom;
• DUNGEON::RoomSelect(&position, f, b) (gcmn 0x005c95c0);
• ChangeScene(-2, -2, -2, -2, f, b).

DUNGEON::RoomSelect:

• It sets roomEnterFlag 1, mapHideFlag 2 (the map painted again but left closed), ccMenu+0x10 (the map status) 3, and level = f.
• The position starts at the room's centre (x, y, 0, 1).
• It then searches the room's anm for OBJ_w_0g10_* (a gate wall), and failing that OBJ_0pae0_* (a door dummy). The first match's world matrix, applied to (0, -200, 0, 1), is the position. Its w is 1.0, which is the heading the arrival takes.
• It also builds an Rz(rotate) matrix and never uses it.
• Areas 71 and 77 have branches of their own (OBJ_user_point in story rooms of types 30 and 31). Those are later volumes' areas, and the port leaves them out.

The port.

• DungeonArea::room_select does the dungeon's part.
• FieldWorld::room_select asks for the change of scene the way a door does.
• The area host's room calls it, then sets the map status to 3.

save_party

Case 88 (0x001af784) sets partyMemberSave to 1 << memberID[s] for each filled slot. The VM already does this. The only host method it needs is Host::party, which both hosts answer from the registry.

M130's block 20 runs at phase 0 in the field. It holds the Skeith entry 7 0 0 0 (which goes into the event manager's entry list for the entry control), battle_ready (already hosted) and save_party. Nothing more was needed.

Checked

• test_dungeon_rt's new test_room_select runs the game's ClearRoom, SetRoom and DUNGEON::RoomSelect in eemu beside the probe's new select request.
  • 208 rooms over the suite's dungeons: 202 in front of a way in and 6 at the centre.
  • It compares the position, the level, roomEnterFlag and the room's hit list (18,947 hit models identical, 60 within the keyed-rotation tolerance).
  • The game's mapHideFlag 2 and map status 3 are asserted.
  • 0 mismatches. The rest of the suite still passes (5 tests).
• piney-game's story_4_item_add_and_room, in story 4's dungeon:
  • add_spc_item gives Orca the item, and declines Sanjuro, who is not built there.
  • room 0 4 builds room 4. The scene changes to it, and Kite arrives where RoomSelect put him.
  • Room 1 would play the story's stream 3 on entry, so the test uses the floor's last room instead.
• item_add_for_a_companion_goes_through_the_menu in the town, after event 2: Orca gets 2 more of a weapon of another job's category in his bag. Piros, who is not in town, is declined. Kite's items are unchanged.