Entry 312

Three reports again: a field's effects table, the Status portraits, a member spoken to while running

2026-09-30uirenderworld

Filescrates/piney-game/src/fx.rscrates/piney-world/src/field_world.rscrates/piney-game/src/session.rscrates/piney-game/src/session/tests/town_return.rsBUGS.md

GitHub issues 2, 3 and 5 (Infection), called fixed too early. The reporter builds from source; the screenshots are from a build after the fixes named (10d3d88, 6edf655, d894a29). None of the three pictures was reproduced. One real fault turned up on the way, in #5's census. Addresses are INF.

Issue 5: what outlives a Data Drain

The report's paralysis is a scroll: Kite's skills (issue 1's screenshot) have no Suvi Lei, and issue 10 is about spell scrolls. The Hanged Man is itemTblD row 37, skill 157, category 11 (tools/battle.py skills: 157 Suvi Lei, paralysis 900 frames on a foe).

Tried on event 3's goblin, each drained once paralysed:

• The Hanged Man from PERSONAL's Items page 11 (drain_with, row 5);
• the same use through FieldWorld::use_item and item_step;
• Suvi Lei cast by Kite (skill_request);
• the drain with and without its movie.

Every time the sparks (two generators of row 198, killFlag 1) went at the drain, with ClearConditionEffect (gcmn 0x00570180). No condition show was lost between the battle and the effects. Nothing followed the drained foe.

The census also held two effects that never ended: 12 (effAfterDrain, ANM_xdhdref0) and 22 (effProtect, ANM_xdhpro*). Both had no object. ccEffectCtrl's constructor (main 0x001c3030) looks each id up in effectTbl (173 rows) when game+0x14, the area, is not 0 (0x001c31bc). Only a Root Town takes effectTbl2 (4 rows). The port's EffectCtrl::new is the town's; AreaFx never switched it. So in every field and dungeon:

• no effect with an object drew (hit rings, the skill circles, protect breaks, the drain's wave);
• the clip-ended ones held their slot for good.

AreaFx::new now sets town false. 12 and 22 end with their clips (a_foe_paralysed_by_the_hanged_man_and_drained_leaves_nothing; [22, 12] still running 120 frames on without the fix). They never drew, so they are not the report's sparks.

FxCensus::char_generators counted only killFlag 0. A condition effect's generators are killFlag 1, so a_drained_paralysed_foe_keeps_no_sparks could not see the sparks it named. It counts both now.

Issue 3: the Status portrait

StatusMenuDisp (gcmn 0x005362d0) walks ccPartyManager's slots to the page and draws that slot's menuFace. SetMenuFace (0x00526940) and the constructor's ccCheckMenuFaceNameParty (0x0056a930) set it. Every row of menuFaceCcsList is a texture of xwin_f00, and all 19 read.

Tried: Orca, and BlackRose with Mistral, called by the Party menu; a field by the console and back by town 0; back by PERSONAL's Gate Out. Each page drew its slot's face with the member's row (the_status_portraits_come_back_from_a_field). A headless GS render of Orca's page before and after the field showed the face both times.

Issue 2: a member spoken to while running

The greeting stopped Mistral within a frame of the press, and she faced Kite, in every state reached:

• to a shop or a landmark (actType 5-10), the stick held through the press or not;
• fresh from the gate;
• after Kite on the CHAT follow order (3, 94, 97; mistral_running_after_kite_stops_when_spoken_to);
• on her way out after Remove, with the registry not full (2).

After event 14 (the story pilot, god on) Mistral is not in Mac Anu at all.

The screenshot's picture needs her in reach but out of view. ccCheckTargetRange (gcmn 0x00519240) takes anyone within 60 plus both widths, whatever the heading. In 2 of 4 menu tries she stood ahead of Kite yet off the screen (calc_tag_pos mode 0).