Entry 306

Outbreak's side events surveyed under the story autopilot

2026-09-30scripttestvolumes

Filescrates/piney-game/src/session/tests/side_events.rscrates/piney-game/src/session/tests/survey.rscrates/piney-game/src/session.rscrates/piney-game/src/start.rscrates/piney-game/src/area.rscrates/piney-event/src/vm/cond.rscrates/piney-event/src/vm/mod.rsplans/outbreak-side-events.mdGAPS.md

Outbreak's side content had never run. tools/evscript.py list on OUT SLUS_205.63 gives it:

• the S3 table, events 250-275 (eventTblS300 at 0x00338000 to eventTblS325 at 0x0033c180; 258 has no script);
• the S4 table's 361 (eventTblS411, 0x00342d20). It opens on 201 and is Outbreak's.

The rest of S4 (350-360) and M4 open on Quarantine's events. No M3 script follows 219. The ML table (401-455) is the members' friendship mails.

That makes 26 events. outbreak_side_event_survey now plays them all; 22 reach their end. plans/outbreak-side-events.md has the table of openers, starts and states.

The survey, by volume

side_survey(disc, cases, frames) serves Infection's side_event_survey and Outbreak's. For each case:

• The start. Each event starts from the story start after its last main opener (crate::start::build).
• Earlier side events. The ones it needs are brought forward as the start brings the story's (Start::bring_forward: ccEventFlagSet, all blocks at level 1). Mutation's are only marked done, as a carried save would hold them, along with the key items its SIGN-02 to SIGN-05 (165-168) give: 287-290, which SEARCH MIMIRU, BEA, KURIMU and A-20 test for.
• Its own earlier blocks. The blocks before the first one set in a field or dungeon (else a town) are marked run. Leading blocks with no settings are left alone, because they play wherever the event is open: 361's block 0 ends it once 264 is done.
• The party. The members the event will not go on without (its not_in_party refusals) are put in the party (start::party_of, the Orca start's code shared).
• The event point. The session is placed at the first located block. In a dungeon it then goes to the event point's room as room_point goes there (FieldWorld::room_select), and the event task is idled at once, as ChangeRequest idles it. The old teleport (Session::go from the harness) left the task at phase 4. A phase-4 block of the new room then played before that room's phase-0 set-up had run: NUKE-2 played its fight scene with no foe entered, and no_active ended it on the spot. Infection's survey had the same flaw.
• The pilot. StoryPilot follows that event alone (Follow::side: story_wants reads it in place of the main story).

What stopped them, and the fixes

• The golden goblins (GOB3-1 to GOB3-4; Infection's 50 and 51 too). The pilot never went after a foe in the open. The new Want::Clear(field) comes from a block that waits on no_active in a field where the event put a foe (entry 5). approach_foe then walks to the live foe with the least HP, sharing approach_bug's path.
  • It leaves live Data Bugs and foes within 300 alone.
  • Without the entry 5 limit it also chased 206's drained Data Bug (field 13, entry 6), and 206's end in the whole run moved from 156,600 to 165,900.
• Near a marker (RACHEL-2, TERASHIMA-1, SIGN-7; Infection's 58 and 61). The new Want::Marker(n) comes from near_marker n wherever distance 0 satisfies it. The pilot walks to event position n in a field, or in the dungeon room that holds it, before the walker, which stands still once it is in its goal room.
• Items. The wants now test has_item with the VM's own rule (piney_event::vm::has_item, made public). RACHEL-2 had waited at point 1 for its last block, which wants item 73; with the test it goes on to trade items 71, 72 and 73 over points 1-4.
• The SEARCH NPCs stand in the Chaos Gate. Marker 0 in a town is markerEvTbl[0], DMY_gate, and ccSelectTarget (INF gcmn 0x00518cc0) lists the gate (its second pass) before PCs (its third). So beside the NPC the pilot lets the stick go and pushes it again, which steps the target on (mode 2: the stick's power rising past 64). This happens only while the gate is the target.
• Fort Ouph. The SEARCH chain opens on 201, but its NPCs go on to Fort Ouph, which only 203 opens (town_move 3), so the chain plays from 204.
• Aid for a lone Kite. With god, a lone Kite in a field or dungeon is raised to LONE_LEVEL (75), as the story does for 206. At 50, SIGN-7's Kite was felled in one blow, a game over near frame 106,000; at 75 it ends at 131,678.

What is left

• GOB3-5 (254). The goblins in field 82 run and heal back, and are not caught in 150,000 frames.
• MARLOWE-2 (256), TERASHIMA-1 (262). The dungeon walk stalls in one room (84's 0-2, 90's 2-6). Its foes stop taking damage while Kite runs about.
• SERVER-3 (263). Black Death (OUT enemyTbl row 176: 9,999 HP, level 70, physical PP defence 9,990, exdefense 2) takes no damage from Kite at 75.
• SEARCH TUKASA (275) ends at the second meeting. Blocks 3 and 11 are both set in Dun Loireag and both hold talked_to 159 in one pass, because operateSet is cleared only at the top of ccThEvent's loop (INF 0x001b5a60) and eventSub (INF 0x001b5ef0) runs every block that holds. Block 11 then ends the event. This follows the documented walk, but it has not been compared with the game running.

No run panicked, left a host call at its default, or faulted.

Checks

• a_followed_side_event_wants_its_foes_down_and_its_marker: Clear(78) for GOB3-1, and Marker(0) and Point(1) for RACHEL-2, only while the pilot follows them.
• search_bt_meets_its_npc_at_six_gates: SEARCH BT (268) plays blocks 2-12 across the six gates and ends.
• gob3_1_golden_goblin_is_run_down: GOB3-1 ends in field 78.
• The whole runs are unchanged: outbreak_whole_story 463,200 frames and mutation_whole_story 613,800. The baseline was measured again from HEAD in a worktree.
• Infection's survey, with god: 50, 51 and 55 end, and 58 and 61 play their near-marker scenes.
• The piney-game suite passes (168) and piney-event's passes. Clippy and fmt are clean.