Entry 238

A party member's skill runs its own animation

2026-09-27battle

Filescrates/piney-battle/src/flow.rscrates/piney-world/src/combat/mod.rscrates/piney-game/src/fx.rscrates/piney-game/src/session.rs

Reported from Mutation's play: Kite's first combat skill from the Skills menu played its sound and did nothing else, and the menu did not let go.

Reproduced

A diagnostic in field 27 walked Kite to the monster portal and, in the fight, chose PERSONAL, Skills, the first skill (Staccatto, 8), and the target. The OK took the SP (100 to 85) and set Kite's skillID to 8. Kite then played his art (act 21, ANM_ctu1ski4) for about 170 frames. The Killer Snaker's HP never moved.

The target menu (65) waits in its tail while the world reports a skill of Kite's running (ccSkillCheck). The run stayed there until the fight ended. It would otherwise have ended at Main's 450-frame limit, for a skill with no system of its own.

The game's rule

_ccSkillRequest (INF gcmn 0x00572c14; MUT 0x005983d4, the same) gives the new ccSkill its own animation (+0x80) when all of these hold:

• the caster casts the skill itself (stype 0);
• the caster is a PC (type bits 0 or 2: *(char) + 8 & 5);
• the skill's row names a file (param + 4);
• xnote_ccs is set.

That animation is new ccAnm, SetAnm(GetChunkAdrsF(GetCCSAdrs(file), "ANM_" file), 0), with ccSkillCheckNote as its note function. xnote_ccs is GetCCSAdrs("xnote") when ccThSkill starts, and XNOTE.CCS is a common (cmn) file, so it is set in every area.

ccSkill::Main (0x005734d4) then works the animation each frame:

• _AnimateForward(frameSpd) while it has data;
• NoteProcess, which runs ccSkillCheckNote:
  • 0x8005 is NoteEventAffect, the hits;
  • 0x8002 is a shock wave;
  • 1 and 2 are the caster's sounds.
• At its end, the caster's skill fields are cleared and the run ends.

The skill files are tiny (CTU0SKI4.CCS is 3,604 bytes): timing and notes only. Nothing draws them.

The port had SkillRun::has_anm and Frame { anim_done, notes } for this, but never set the first, and the combat loop always handed main an empty frame.

The fix

• The run (Skills::add). A run gets has_anm by the rule above, with the new Skills::xnote in for xnote_ccs. The combat sets it on; harnesses without the file leave it off, as before.
• The effects side (FxTasks::skill_anim, in AreaFx). On a run's first call it reads the skill's file once and makes a Play of ANM_<file>. Each call then steps it (forward_notes: its end and its notes). The Play goes with the run (spell_remove).
• The combat loop. It steps a run's animation where Main does and hands the notes and the end to SkillRun::main. If the file or the animation cannot be had, the run ends at once. The game's would never end.

Checked

The new test mutation_skill_from_the_menus plays the reported steps:

• the menu shuts within 400 frames of the OK;
• the monster takes damage.

In the diagnostic run, the monster fell from 1,250 HP in a string of Staccatto's hits, the skill ended with its animation (about 170 frames), and the menu shut.

The Rust suites of piney-battle, piney-world and piney-game pass. So do test_battle_flow_rs, test_battle_kite_rs, test_battle_party_ai_rs and test_battle_rs on Infection and Mutation.

Two other reports looked at here:

• CHAT after a fight. Mutation's ChatMenu is Infection's instruction for instruction. In a field or dungeon it always has the Skill Usage, Strategy and Members pages, in battle or not. The one extra order, the Sprite Ocarina's, shows only in a dungeon out of battle. The battle mode itself ends properly (in_battle 1, 2, then 0).
• The quit prompt over another menu drew its text into that menu's rows. It kept a held picture without its uploads, while a frame's upload ids restart at 0. It now sends the picture's uploads again each frame (c3a70e3).