ba.groundbind "目标变为地面单位，无法移动；使部分近地悬浮敌人掉落；对重量大于3的单位持续时间减半" (予愿安洁莉娜 S2 needs it):
- buffs.js: STATUS groundbind (flag groundbind + noMove, move x0), FLAG_KEYS, RESIST_STATUSES (as 束缚, [ASSUMED]);
- units.js isFlying: an enemy under 缚地 is a ground unit unless a 浮空 lifts it (same result as before without the flag);
- battle/status.js: 浮空 lands on a 缚地 data flyer (PRTS 异常效果 "…且不持有缚地异常…则Buff取消"), 缚地 halved on weight > 3;
- content/enemies: 掠海漂移体 / 吉兆飞鳞 drop on 缚地 as their PRTS text says (吉兆飞鳞 stays down while it holds);
- render/style.js: the 束缚 icon for it; docs/SIM.md (§2 isFlying, §3 flags / status table / 抵抗).
Nothing applies it yet: every golden family is unchanged. Test: test/sim/feedback5-groundbind.test.js.
Conflicts: server/match/bot.js (W4A's comment + W5K's player data view pgd), test/golden/diy.json (regenerated). data/ equals an offline build of the merged tools/build-data.mjs. Golden regenerated on the merged code: against feedback5 the moved roster / bonds / fields / matches / standins scenarios are exactly W5M's per-PR attribution (#160 skill events, #109 deploy-timed lifecycle, #66 灵巧 sources and hidden layers, 83fc408 歌蕾蒂娅 placement), plus coop2-NORMAL-70-diy (fields operators those PRs touch) and diy-051 / diy-052 (events.skill only: Misery's deploy passives, #160); against fb5-merge4 only feedback5's own additions (the diy kits, coop2-NORMAL-70-diy).
Bonds of 自选 pieces checked everywhere membership is decided (the owner's reminder of 2026-10-05): the match's counts through the player's data view (煌 in 炎 and 维多利亚, one member per slot, 绝技 elites), the sim's bond effects through def.bonds (叙拉古 on deployment, 协防 ×1.2 / elite ×1.4), the client's strip, popups, cards and detail card. Golden unchanged.
Tests: test/ui/diy-ui.test.js (15), test/match/diy-bonds.test.js (4), test/i18n.test.js pins screens/diy.js; test/ui/diy.e2e.test.js (SP_E2E=1, run once: pass).
Ported by hand:
- #109 kit hunks → kits/ops (the tier files stay re-export shims): 宴 chess_char_1_18-utage.js, 野鬃
chess_char_1_19-wildmn.js, 砾 chess_char_2_12-gravel.js, 斯卡蒂 chess_char_3_05-skadi.js, 伊内丝
chess_char_4_04-ines.js, 缄默德克萨斯 chess_char_4_16-texas2.js, 耀骑士临光 chess_char_6_17-nearl2.js. The +/- lines
are identical to #109's tier1/2/3/4/6 diffs.
- #109 generic.js against W5D's own-duration passives (一击即退): the deployment window is computed where #109 computes
it, with W5D's case included. Both forms run #109's duration lifecycle (activateOnDeploy, spType 'none', the
skill's own mods), and W5D's 冲锋号令 DP start effect is kept.
- #103 room.js through i18n phase 1: t('{name}邀请你加入卫戍协议：盟约【{difficulty}】') with the translated difficulty
name (the Chinese text is the PR's), plus its English line in public/i18n/en.json.
- data/assets.json: feedback5's manifest plus #72's audio.bgm.combatAlts (after boss, the plan order). files +3 and
bytes +5 037 595 as measured on feedback4; the hash is recomputed with the generator's contentHash. No other data/
file changed.
- docs: DESIGN §5.6 (#109's timed deployment paragraph before feedback5's trigger rules; "Permanent passive skills are
always on"), SIM §7.4 (both generic passive cases).
- Auto-merged and checked: #149 bond-bar fold (still in screens/game.js), #160 skills.js / spine.js, #109 skills.js /
snapshot.js, #66 addon/battle.js (restTile / isDown live in sim/battle/tiles.js), #150 units.js, #72 audio.js /
plan.mjs, #74 plan.mjs / docs.
Where the PRs met feedback5 work:
- 突袭 (W5L, GitHub #49): raidPoll counts a skill that needs no SP while it is on. That is a passive, as before, or a
#109 deploy-timed skill while its window runs. Once the window has ended only the idle trigger is left. Without this
change 缄默德克萨斯 / 宴 lost #49. research 02 §3.18 says so.
- Tests changed to #109's representation: bonds_addon #49 test (kind 'duration', activateOnDeploy) plus a new test
(window over → idle trigger only), and standin.test.js's 预备干员 PASSIVE branch (the skill's mods instead of the
generic:passive buff).
Golden regenerated on the merged code. Attribution: each PR's gameplay change was reverted alone in a scratch copy, and
all three together. With all three reverted the digests equal feedback5's exactly. Every moved scenario was replayed
and fields the unit that explains it.
- #160 only (0.5 s skill events of deploy passives; events.skill only): roster-001/006/025/049, bond-arcaneShip-low,
bond-emptyShip-high, bond-investShip-high, bond-kjeragShip-low, bond-siracusaShip-low/high, bond-victoriaShip-low,
bond-yanShip-high, boss-boss_2/4/6-solo, hidden-boss_8/10-solo, standin-03/05/08/10, diy-005/006 (普罗旺斯 S1,
琳琅诗怀雅 S1/S2, 星熊 S2, 凯瑟琳 S1, Misery S3 as stand-in / 自选 prototype).
- #109 only (deploy-timed skill state and lifecycle; 突袭 技能就绪 limited to the window; generic 一击即退):
roster-003/004/005/021/023/024/039/041/042/046/047, bond-egirShip-high, bond-indomShip-high,
bond-kazimierzShip-low/high, hidden-boss_8-pair, standin-01, coop4-FUNNY-6, coop2-NORMAL-8-serverrun.
- #66 only (灵巧's aura stays only for a killed member, from its body tile): roster-028, coop2-NORMAL-3,
coop3-ABYSS-5, coop4-ABYSS-7, solo-HARD-9-boosted.
- #109 + #160: roster-002, roster-029, bond-investShip-low, bond-raidShip-low/high, bond-suntShip-low,
bond-visiShip-high.
- #66 + #160: roster-008, roster-027.
- #66 + #109: roster-022, roster-040, bond-indomShip-low, boss-boss_2/4/6-pair, hidden-boss_10-pair, standin-06,
coop2-HARD-4, coop2-NORMAL-14-standins.
- all three: roster-010, roster-012, roster-031.
69 scenarios. Every other digest is unchanged.
The scouted prep board carries hand and 临时整备区 items as units (#129). The frame loop syncs every unit of the snapshot and ItemView had no sync, so it threw on every frame and the teammate's operators were never drawn. Contributed by @2321Robin.
PIXI loading, pick records, tunables, view kind, unit info and the asset / GL
host now live in public/js/render/app/*.js. createFieldView stays in app.js,
unchanged, and every name the file used to export is re-exported. One
pixiPromise. Two source pins follow the lines that moved.
public/js/render/fx.js becomes a re-export barrel. Speeds, camera heights, fonts, the kind table and the shared tunables move verbatim into public/js/render/fx/*.js. FxSystem keeps its constructor, clear, update, counts and destroy; the other methods move verbatim into container classes installed on FxSystem.prototype in source order. simSpeeds and fontsReady each stay one instance. The dynamic import of /sim/constants.js stays next to setSimProjectileSpeeds.
The prep scout of another player's board carries their 整备区 hand and 临时整备区 as units on the bench rows (like the own prep bench), the equipped items of their operators, their strategy in the drawer and their effects column (the official mode shows a teammate's hand too). The scout signature includes the hand / temp slots, so a moved card refreshes the watcher. Closes#128. Contributed by @2321Robin.
The RES bonus was already off. The enemy card now greys a SILENCE-format line from the live silenced flag, and the status row leaves the 折射 icon out while silence is up.
131 of the 135 Back skeletons have no Die clip, so SpineActor.die() changed nothing and the
attack / skill / idle loop went on under the redeploy ring. UnitView._wantsBack() now wants the
Back model only while the unit stands (or when its Back skeleton has a Die clip of its own);
die() / revive() mark the view and its update swaps the model in place after the frame's events
(a death and a redeploy in one frame load nothing). The Front model replays the fall (held end
for a unit already down), gets the knocked-down grey, and the redeploy brings the Back model back
with the deploy clip carried over. A skeleton without a Die clip holds its idle's first frame.
A reload (a phone browser or Chrome's Memory Saver discarding the background tab) made the asset store download
/data/assets.json a second time; createFieldView waited at most 4 s for it, every view looked its model and avatar up
once, and a failed download kept an empty manifest for the page. Now the field host seeds the store with the game
data's copies (seedAssets); without one, ready() retries like data.js, never keeps a failure and tries again with a
bounded backoff, and onChange makes every view re-resolve (UnitView.retryAssets, ItemView.setIcon). A failed or
timed-out Spine load is retried by its view (SPINE_RETRY_MS, acquire { retry }) and at once when the tab is shown.
Battle mode holds the Spine cache (spine.hold()), so a battle that began in a hidden tab keeps its models; the idle
budget still applies and the quiet budget still frees everything in the lobby, room and result.
While the own prep board is shown and the shop bar is folded, the game asks for the prep camera with shop: false
(left_prepare_camera_param / the boss prepare params) and the view keeps the bench above the folded shop's HUD band:
the higher of the folded tab and the corner buttons' hit areas (hudBands(kind, size, { shop: false }), cornerBand).
Unfolding returns to the shop camera. Never over the pen, a teammate's board or a battle; deferred while a piece is
dragged or its direction is chosen (gameLogic prepCameraFor / foldCamera). Below 768 px of width the corner stays on
one row while the bar is folded. Tile px at row 10: 1920x1080 109.8 -> 130.7, 844x390 35.9 -> 44.3,
756x366 31.4 -> 40.1, every bench pad fully pressable.
- render/interp.js isCosmeticEvent: an fx with a form is state, never dropped as stale nor shed from a full queue;
a late one switches the model without its telegraph, its closing clip shortened (render/app.js)
- battle/runner.js: a backlog past HELD_MAX keeps only the last status / skill toggle per unit; a battle that ended
while the tab was hidden delivers its backlog and last snapshot when the tab is shown
- screens/game.js: a new field meta for the field on screen buffers its frames until the view re-enters it
- browser test stalls the main thread (blocking 1 s long tasks); Node tests drive the runner into the render buffer
render/interp.js keeps the body tile of a b.snap `down` entry (row, col; only
when both are integers, so a 4-element entry is unchanged) and
render/units.js setDown moves the knocked-down view and its ring there: an
operator that fell on another piece's home lies on its own home (PRTS
卫戍协议/帮助 "…尝试返回其自身的初始位置"), and a view built from the unit's
deploy info (a field joined mid-battle) no longer stays on its first tile.
Tests: test/render/feedback1f-down-tile.test.js (fake PIXI, the sim's own
snapshot); opt-in test/render/feedback1f-down-tile.browser.test.js (real sim
frames in headless Chrome: drawn, picked and redeployed on its own home).
Review of the E3 fix (阵法术师 / 轰击术师 strike every enemy in range):
- 卡涅利安 S1 沙暴守卫 charged: PRTS 备注 "应用蓄力时：应用技能未开启时的特性"
keeps the whole skill-off trait in force, 不攻击 included. A kit trait
canAttack now stops her attacks while a charged S1 runs; the guard stays.
The E3 test casts S1 uncharged and a new test pins both casts.
- 卡涅利安 S3 食噬之印 charged: the mark stacks before the attack's damage
(PRTS 备注 "于攻击造成伤害前生效"): x1.2 on the first hit, x2.0 from the 5th
(was one attack late). One mark per enemy: another 卡涅利安 only adds
stacks, the bonus stays the setter's (PRTS 备注 popup).
- rangeAoe is applied after every override (kit trait included) and sets
allInRange itself; SUB.phalanx / blastcaster carry rangeAoe only.
- 流形 copies (tokens.js and the 缪尔赛思 kit) turn a copied 'beam' into a
plain bolt: the copy takes no attack shape.
- fx.js: the dead blastcaster shell colour is gone (and its test).
- SIM.md / comments cite PRTS 溅射半径一览 (no splash row for these
branches), note that stealthed enemies are not struck unless revealed or
blocked, and list the table's other radii as open deviations.
data/assets.json pointed 灼热源石虫 / 炽焰源石虫 at /assets/local/ files that only exist
where tools/local-extract ran, with no fallback: without the extraction the
asset tests failed, setup re-ran fetch-assets (rewriting the tracked manifest)
and the slugs drew as icon diamonds.
- The web model stays the 源石虫 alias (spine + spineAliasOf). The official
model is enemies[id].spineLocal: file names in the data/local-assets.json
group spine/enemy/<id> plus the parsed metadata, from the committed
tools/assets/local-enemy-spines.json (fetch-assets --local-spines refreshes
it from the extracted files, read only). The manifest is byte-identical with
or without the extraction and has no /assets/local/ URL.
- Client: assets.js spineEntry uses spineLocal only when the local manifest
lists every file (createFieldView awaits assets.local() with the manifest);
UnitView loads the web model when the local one fails.
- extract.py writes the atlas with size: / pma: true itself (the same text as
atlas.mjs normalizeAtlas), so the extraction is usable without a fetch pass.
- setup notes when an older extraction lacks the enemy models.
Community report E3 "卡涅利安的攻击不是真群攻" was real: the phalanx trait
"技能开启时攻击造成群体法术伤害" was read as one bolt at the priority target
plus a 1.1-tile splash, so spread enemies took one hit per attack (act2 m01,
round 10: 2.3 of 12.4 enemies on her range per attack, never all).
Officially a 阵法术师 attack hits every enemy inside the attack range at
once, the same damage near and far (Arknights Terra Wiki, Phalanx Caster;
PRTS 林 S3 note; PRTS 作战机制 AOE), and so does the 轰击术师 line
(Terra Wiki, Blast Caster; PRTS 作战机制: 伊芙利特's 炎爆 is a range-locked
AoE). The 扩散术师 keep their 1.1-tile splash around the target.
- professions.js: phalanx and blastcaster get allInRange + rangeAoe
(instant 'beam' hits, no projectile) instead of the 1.1 splash.
- 卡涅利安 S3 食噬之印: ATK climbs in 1 s steps to the full bonus at 20 s
(PRTS note) instead of continuously over the 21 s duration.
- fx.js: phalanx / blastcaster leave the splash-ring list.
- Tests: real battles counting damaged enemies per attack for every
phalanx skill, both blast casters, a splash-caster guard, a real stage
and wave, the S3 ramp; opt-in headless-Chrome check of the strikes.
- SIM.md profession rows and kit comments updated.
Every setForm fx now carries 'form' (shared/protocol.js fxForm). The runner keeps
these fx in catch-up frames (keepsState), holds the state-bearing events of the
battle on screen while the tab is hidden and delivers them with the first frame
back (re-entering from the field meta past HELD_MAX); screens/game.js buffers
them before a field is entered. A husk beaten while its stand-up clip plays dies
on the husk's clip. 幽灵 form cited from PRTS; 特战术师 flyer question recorded.
The bomb-less mode lived only in the client that saw the fx 'phase' event. UnitInfo now
carries an enemy's recorded mode (unit.form, set by kitBombd at the release), and the
renderer keeps it on the unit info, so a field watched or rejoined later starts on the
*_2 clips.
The drone triggered its one bomb, but the damage and the blast landed in the trigger tick
while the model kept the bomb hanging under it for good. Official (PRTS 暴鸰; the client's
battle prefab enemy_1040_bombd and projectile_bombd): the cast plays the Attack clip, the bomb
leaves on its OnAttack event (0.267 s), flies to the target at 5 tiles/s and explodes on it
and the 8 tiles around; the drone switches to its bomb-less clips and speeds up.
Sim (kitBombd): release after BOMBD_RELEASE with an 'atk' event of kind 'droneBomb' and fx
'phase' {kind: 'bombed'}, a homing projectile (PROJECTILE_SPEEDS.droneBomb 5) whose arrival
deals the damage, then the speed buff. Render: FORMS.enemy_1040_bombd.bombed (*_2 clips),
PROJ.droneBomb (a falling shell), a one-off cast plays the attack clip once at its own
speed instead of the drone's 5 s attack rhythm, no puff for the silent mode change.
sganggs