Follow-up to the multi-instancing migration. Wraps the process-shared engine
statics that aren't ambient-fronted but race between concurrent battles:
- UnityEngine.Resources._loaded: Dictionary -> ConcurrentDictionary.GetOrAdd
(the shared prefab cache keyed by path; concurrent first-misses produced
duplicate GameObjects + Dictionary corruption)
- UnityEngine.GameObject._components: Dictionary -> ConcurrentDictionary with
Interlocked.CompareExchange init (Resources.Load returns SHARED prefab
GameObjects, so two engines' Setup() can race on the same _components map
— surfaced as "Operations that change non-concurrent collections" crashes
during BattleManagerBase ctor's GetComponent<T>() chain)
- Wizard.LocalLog: single static lock around all mutating entry points
(StringBuilder _lastTraceLogStringBuilder + ~12 mutable string/bool/int
scratch fields; serializing the trace-log surface is cheap since logging
is not the hot path)
Flips SVSim.BattleEngine.Tests assembly Parallelizable scope from Self to
Fixtures and restructures MultiInstanceEngineTests.StressN_BaselineMatches so
Setup runs INSIDE Task.Run (was previously serialized as a workaround for the
LocalLog races). The fixture is also lifted to ParallelScope.All so the
two-engines and stress tests can run alongside each other.
Suite fully green under fixture parallelism (59/0/2 across 3 consecutive runs);
SVSim.UnitTests still 1054/0/0 — true multi-instance correctness is now proved
end-to-end in tests rather than gated behind a serial workaround.
Manifest sha refresh + new patch artifact for the LocalLog edit (decomp-origin);
the two shim files are authored, so no metadata update is needed for them.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Replace trivially-true Pp>=0 with concrete post-Setup pins (LeaderLife=20,
Pp=0, HandCount=3). Drop the unused seed parameter from SampleDeck - every
call already returned the same vanilla deck, and the StressN test name 'Random
Decks' overpromised. The cross-contamination property the test pins (parallel
LeaderLife[] equals sequential LeaderLife[]) holds with identical decks +
distinct masterSeeds, which is what's actually being verified.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Step 6 of multi-instancing migration. HeadlessEngineEnv.EnsureInitialized
is split into EnsureProcessGlobals (idempotent, process-once) +
SeedCharaIdsOnCurrentAmbient (per-test). New TestBattleScope IDisposable
sets up a fresh BattleAmbientContext per test. NonParallelizable removed
from converted classes; assembly-level Parallelizable(Fixtures) enabled.
SVSim.BattleEngine.Tests fully green under parallel test execution.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Step 5 of multi-instancing migration. GameMgr.GetIns() now resolves through
BattleAmbient.Require() (throws when no scope active — fail-fast since engine
callers unconditionally dereference). SessionBattleEngine now owns a single
BattleAmbientContext, pushed via BattleAmbient.Enter at Setup/Receive/all
~30 read accessors and Debug* seams.
EngineGlobalInit.WirePerSessionGameMgr extracted out of the _done-gated block:
GameMgr is now per-session (ctx.GameMgr is a fresh `new()` per SessionBattleEngine),
so the DataMgr chara ids + NetworkUserInfoData seeding must run every Setup, not
process-once. The wiring itself is already idempotent. Without this, second-or-
later sessions in a process NRE in NetworkBattleManagerBase.CreateBackgroundId.
Expected state: SVSim.BattleEngine.Tests have known-failing tests that don't
go through SessionBattleEngine (Task 6 wraps HeadlessFixture). SVSim.UnitTests
mostly recover; residual failures (deal-frame Accepted:false in conductor
integration tests) are captured in
data_dumps/task5-test-output/failing-tests-after-task5-node-postwrap.txt for
Task 7.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Setter is the asymmetric one (write-through inside scope, unlike ViewerId's
no-op-in-scope) — adding parity with the SetRealTimeNetworkBattle ambient
setter test to catch future regressions if the routing branch is touched.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Step 4 of multi-instancing migration. Three additional per-battle statics
front-fronted by BattleAmbient.Current, each with a static fallback for
unwrapped callers. ViewerId's SavedataManager-persisting setter is preserved
on the fallback path; inside a scope, the setter is a no-op (the per-battle
perspective is fixed at scope entry).
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Step 3 of multi-instancing migration. The dominant per-battle singleton now
resolves through BattleAmbient.Current.Mgr when a scope is active. The legacy
'main' field is renamed _mainFallback and retained for unwrapped callers
(tests, anything not yet scope-wrapped). GetIns() still returns null when
neither is set, preserving the '?.Foo ?? default' patterns in engine code.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Step 2 of multi-instancing migration. Both flags now resolve through
BattleAmbient.Current when a scope is active, otherwise hit a static fallback
that preserves today's behavior unchanged for unwrapped callers.
Suite green: SVSim.BattleEngine.Tests pass; SVSim.UnitTests baseline holds.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Step 1 of the engine multi-instancing migration. Standalone infrastructure;
no engine static reads/writes through it yet. Scope is reentrant (restores
prior on dispose), AsyncLocal flows across awaits, and isolated between
concurrent Task.Run flows.
See docs/superpowers/specs/2026-06-07-engine-multi-instancing-design.md.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Six distinct fixes accumulated over live-test iterations against four bids
(654473755566, 806245601092, 283192092460, 131549100204, 799755786270) — together
they take the shadow engine from "throws on the first non-mulligan play" to
"survives a full PvP battle, only weird-edge-case Unity touches still left to whack".
1. Engine StableRandom seed aligned with clients' Matched.seed
(BattleSession.EnsureEngineSetup, NodeNativeBattleHarness.Create). Clients seed
_stableRandom with BattleSeeds.Stable(masterSeed) (the value the node ships in
Matched.seed); we were passing the RAW masterSeed to engine.Setup, so every
StableRandom call diverged from call #1 onward — every turn-1+ draw picked a
different deck position than the clients. Verified Stable(1184631275)=1543475792
matches the wire on bid 654473755566.
2. SeedDeck advances cardTotalNum to deck.Count+1 + pins BattleStartDeckCardList.
Mirrors SBattleLoad.InitPlayer's tail (SBattleLoad.cs:1292). Without it,
skill-generated tokens auto-assigned Index 0,1,... and COLLIDED with deck-loaded
indices 1..40 — silent until something addressed the deck card with the
colliding Index (Hoverboarder at deck idx 1 + a token at engine Index 1 made
GetBattleCardIdx's SingleOrDefault throw on bid 806245601092).
3. BattleCardView.GameObject lazily non-null in the shim (ViewUiTouchStubs.cs).
The IsRecovery card-create delegate (NetworkBattleManagerBase.cs:379) passes
null cardGameObject; Skill_metamorphose.cs:147 in the in-play branch then NRE'd
on `metamorphosedCard.BattleCardView.GameObject.transform.rotation = identity`,
a purely cosmetic touch with no game-state implication. Bid 283192092460:
Petrification on a board follower.
4. TranslateChoiceKeyAction unwraps wrapped selectCard on shadow ingest
(SessionBattleEngine.cs, sibling to TranslateTargetOwners). Live sender-send
wires Choice plays as selectCard:{cardId:[...], open:0}; engine's
ConvertToListInt does `value as List<object>` — a Dict casts to null and
foreach NREs. The receiver's swallow-all catch (NetworkBattleReceiver.cs:1255)
logs to Debug.LogError + LocalLog — both shimmed/no-op'd headlessly — and
returns false, but Receive calls ReceivedMessage with checkBreakData:false so
the false isn't propagated. The play continues with choiceIdList=[], the chosen
branch never resolves, the played card stays in hand; a later targeted play
(A's bounce on B's "board" idx 20) then can't find the target → NRE on null in
ActionProcessor.PlayCard:407. Bid 131549100204: B's Resonance + A's bounce.
Opponent-relay path is unaffected — node strips selectCard from broadcasts.
5. HeadlessHandViewStub overrides HandUnfocus/HandFocus/FocusRearrangeHandHand
to return NullVfx. CreateHandControl returns null in headless; the base
methods unconditionally deref `_handControl.SetHandState(...)`. A follower
with a when_spell_play Heal trigger fired on its leader for amount 0 — even
a 0-heal drives ApplyHealing → CreatePullHandInVfx → HandUnfocus → NRE.
Bid 799755786270: two consecutive spell plays both crashed this stack.
Added InternalsVisibleTo("SVSim.BattleEngine.Tests") so the shim-level
regression tests can pin the no-op contracts directly.
Plus the previous-session fixes carried in this same uncommitted state
(see docs/superpowers/plans/2026-06-07-shadow-engine-desync-handoff.md):
- doesPlayerGoFirst:true + mgr.IsFirst:true (turn-1 draw count correct
per seat)
- RecoveryOperationCollection.PlayHandCardOperation routes all type:30
through PlaySkillSelectHandCardOperation (skips the two-phase user-select
guard that aborts targeted spells in recovery)
- ShadowFeed + ToRawBody: server-generated typed bodies (DealBody, etc.)
converted to RawBody before engine.Receive (`env.Body as RawBody`
returned null for typed bodies)
- Ready idxChangeSeed seeds A's XorShift via the receiver; B's seed is
injected via SeedOppoIdxChange (BattleSeeds.IdxChange + viewerId)
- ReadySpin defaulted to 0 (was 243) — non-zero double-cranks the shadow
which ingests BOTH sides' Ready frames on one stream
Test counts: SVSim.UnitTests 1054/1054, SVSim.BattleEngine.Tests 34/34.
Open: known-residual Unity touches are individual whack-a-mole now (per-card
skill edge cases), not the structural divergences fixed here.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
The engine's receive CONDUCTOR fuses each authoritative mutation behind a view
call: the play mutation is an InstantVfx registered to VfxMgr, and the deal hand
is seated by MulliganPhaseBase.StartDeal wired to OperateReceive.OnReceiveDeal.
Headless, the shared VfxMgr no-op'd registration (correct for the direct
ActionProcessor path the M2-M12 oracles use) and OnReceiveDeal was never wired,
so the receive path resolved nothing.
Untangle (Candidate B, zero Engine logic edits):
- InstantVfx.Run() opt-in executor (authored shim).
- HeadlessConductorVfxMgr : VfxMgr runs registered InstantVfx; wired only via the
node's SessionContentsCreator.CreateVfxMgr (verified the receive mgr's VfxMgr
comes from there — BattleManagerBase.cs:768). M2-M12 use HeadlessContentsCreator,
so they're isolated by construction.
- WireMulliganPhase: construct NetworkMulliganPhase + MulliganEventSetting() to
install OnReceiveDeal -> StartDeal (the node never pumps the phase machine).
View no-op surface (the 7 from the probe, minus 1 not hit; +1 emergent):
- Deal wiring (NetworkMulliganPhase) [node seed]
- MulliganInfoControl._partsPlayer/_partsOpponent._exchangeMark/_keepZone/_abandonZone [node seed: prefab + SeedMulliganInfoControl]
- Data.BattleRecoveryInfo (IsMulliganEnd=false) [EngineGlobalInit seed]
- IBattlePlayerView.PlayQueueView -> HeadlessPlayQueueViewStub [_IfaceImpl.g.cs, both getters]
- DetailMgr.DetailPanelControl/SubDetailPanelControl [node seed]
- BattleCardIconAnimations.collection (emergent: UpdateInPlayBattleCardIconLabel) -> HeadlessIconAnimations empty SkillCollectionBase [_IfaceImpl.g.cs]
- BattleMenuBtn (probe item 7): NOT hit on the vanilla path; not seeded.
Oracle (HeadlessConductorTests): node Deal seats 3-card hand; a vanilla
hand-card Play leaves hand (-1), adds board (+1), drops PP by cost.
Regression: 24/24 BattleEngine.Tests oracles (M2-M12) green; 241/241
SVSim.UnitTests BattleNode green. The 2 SessionEngine capture-replay shadow
tests are marked Ignore (superseded): they passed VACUOUSLY when the receive
path resolved nothing; with resolution live they hit the documented
capture-replay draw-misalignment artifact. Node-native battles are the oracle.
Drift: no drift.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Adds SessionBattleEngine.PlayedCardSpellboost + PeekPlayedCardSpellboost (pre-resolve
read of the acting seat's hand card by Index==playIdx) and a CaptureReplay.InterleavedSends
helper. The non-circular capture oracle (engine-derived spellboost vs prod's independent
emission to cl2: idx2->1, idx14->2) is added but [Ignore]'d: the headless receive path does
not apply the wire's authoritative orderList (Deal/Swap don't seat the mulligan hand, draws
follow the seeded deck top instead of the wire move ops, plays never remove the card, alter
spellboost never accumulates), so the engine cannot yet DERIVE the count. Closing this needs
an Engine/*.cs + VfxMgr-execution logic change (escalation per the N2 playbook), not a
mechanical no-op fill. Read surface, node + engine builds, drift, and the rest of the
SessionEngine suite are green.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Full single-client capture replay (cl1 send=player seat, receive=opponent seat,
ts-ordered) ingests end-to-end: 33 frames, 0 rejects, 0 invariant violations at
turn boundaries (leader life/PP/board/hand).
Headless gaps filled per playbook (no Engine/ drift):
- IsRecovery=true after construction: the engine's own headless replay mode gates
the live view/UI layer off (BattleUIContainer, turn-control UI, VFX waits) while
keeping the live NetworkBattleReceiver (ND4) and authoritative state.
- Seed ToolboxGame.RealTimeNetworkAgent, BattleUIContainer, _backGround, and
per-player NullPlayerEmotion no-ops the receive/turn cycle dereferences.
- _IfaceImpl.g.cs (shim, not Engine/): BattleCardView.BattleCardIconAnimations
returns a lazy non-null no-op so the opponent card-reveal icon-init (deferred
VFX) doesn't NRE.
- HeadlessCardMaster.Load made cumulative: it replaced the global CardMaster each
call, so a Load(deck) evicted the oracle card set and broke tests run after.
Adds board-state accessors (LeaderLife/Pp/HandCount/BoardCount) and CaptureReplay
ts ordering.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Receive feeds the decoded frame into the mgr's own NetworkBattleReceiver
(isHaveSequence:true, checkBreakData:false — mirroring the engine's
RecoveryDataHandler frame replay), reboxing object?->object for nested data.
No engine gaps surfaced; the only fix was a test-harness one (load all deck ids
in a single HeadlessCardMaster.Load — per-id calls each replace the master).
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
CaptureReplay normalizes the capture's send/receive envelope asymmetry (send
frames carry uri at top level + bare payload body; receive frames carry a full
envelope body) and extracts selfDeck + master seed from Matched.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Mirrors HeadlessFixture.NewNetworkEmitBattle wiring (opponent seating, leader
life, card templates, deck seeding) minus the emit-only RealTimeNetworkAgent
scaffolding (shadow only receives). Probe passed first run — M13 already filled
the network-mgr construction gaps. No Engine/ edits; drift clean.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
SessionContentsCreator mirrors the test HeadlessContentsCreator fully (all
IBattleMgrContentsCreator members) so it compiles; Setup/Receive throw pending
the Task 3/4 probes. New files use the 'engine' extern alias.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Card 103111050 (ELF cost-1 self-buff follower) carries skill_condition
`character=me&target=self&play_count>2`. New GatedConditionalOracleTests asserts
BOTH branches of the SAME card in one fixture, varying only the seeded per-turn
play count via the public AddCurrentTrunPlayCount seam (M4/M10):
* gate TRUE (seed 5 > 2) -> when_play powerup fires -> 1/1 -> 2/2
* gate FALSE (seed 0 <= 2) -> powerup is a NO-OP (stays 1/1), BUT the card
still pays its cost and still moves hand -> board.
This proves the engine SUPPRESSES an effect when a skill_condition is false (the
dual of "effect fires" — no prior milestone proved this), and that the gate
suppresses the EFFECT, not the PLAY. Jointly satisfiable only by a correctly-
gating engine: an always-buffs engine fails FALSE, a never-buffs engine fails
TRUE. Reuses the M4-proven buff dimension so the only new thing under test is the
conditional itself.
11/11 green; engine 0 errors; check_drift clean; ZERO new Engine copies / ZERO
shim / ZERO manifest changes — a clean milestone like M4/M6/M8/M10 (condition
evaluation is pure logic on copied engine code).
Load-bearing proof (M4/M6/M8/M10 discipline; the test passed on its first run,
which proves nothing alone): swapped the two seeds -> exactly the 4 stat
assertions failed for the right reason (formerly-TRUE branch seeded 0 took no
buff [1/1, expected 2/2]; formerly-FALSE branch seeded 5 buffed [2/2, expected
1/1]), while the cost-paid + hand->board assertions stayed green in both branches
— confirming the gate drives ONLY the effect and the card resolves regardless.
Reverted -> 11/11.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
DynamicValueSpellOracleTests proves the engine COMPUTES an effect magnitude
from live game state (the value the wire can't carry). Card 112134010's
`when_play damage={me.play_count}-1` resolves via the proven IsForecast/
IsRecovery + ActionProcessor.PlayCard (DP4) path; the oracle asserts the
damage equals the engine's own live GetCurrentTurnPlayCount() - 1, not a
literal. Seeds play_count via M4's AddCurrentTrunPlayCount seam; lone
surviving enemy 13/13 gives a clean life-delta; selectedCards: null
(auto-target AoE). 10/10 green; zero Engine/shim/manifest changes; drift
clean. First-unknown resolved by the first RED: the per-play +1 lives in
OnBeforePlayCard (wired only via OperateMgr/Prediction), so the direct-
ActionProcessor harness reads exactly the seeded count (damage == seeded-1);
load-bearing proven by varying the seed 4->7 and watching damage track 3->6.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Proves the deck->hand transfer dimension (design §5 draw oracle) — the last
deterministic, non-RNG card-effect class no prior milestone touched (M3/M4/M6/M8
moved stats, M2/M5/M7 the board, M3 the leader).
Card 800114010 (clan-1 ELF cost-1 when_play draw 1 from own deck, ungated, no
evo/preprocess). The resume-guide's skill_target=none/no-RNG shape does not exist
in cards.json — EVERY draw selects from the deck via a random_count filter
(skill_option is always literally 'none'). RNG neutralized structurally: seed the
deck with EXACTLY ONE known card so random_count=1 is deterministic regardless of
seed. New primitive HeadlessEngineEnv.SeedDeck (create via the null-view seam +
engine AddToDeck). Oracle DrawSpellOracleTests asserts: seeded card moves deck->hand
(by id + by reference), deck -1, drawn card IsInHand, spell pays cost + leaves hand
+ resolves to cemetery, board/opponent untouched. Load-bearing confirmed the M7 way
(seed a different id -> the by-id assertion fails).
Shim gap fixed (the predicted M9 cost): Skill_draw's BattleLog tail
(UpdateFusionedCardSkillDrewCard, unguarded; + the IsBattleLog AddLogSkillDrawCard
calls) dereferences BattleLogManager.GetInstance(), an M1 'default!' null singleton
-> NRE after the draw already committed. One-line HEADLESS-FIX (M9) in
BattleLogManager.g.cs returns the existing _instance singleton (all its methods are
no-ops), per the M2/M7 Null*-singleton playbook. No Engine/ edit (drift clean).
9/9 green; check_drift.py clean; engine still 0 Error(s).
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
A when_play damage=5 spell (the M6 card 800134020) played at a select_count=1
enemy follower with life <= 5 kills it as a consequence of damage -> life <= 0 ->
the dead-check + the same RemoveInplayCard/cemetery path M7 lit up (the dominant
real-card removal mechanic), reached through combat math rather than `destroy`.
Oracle LethalDamageSpellOracleTests: selected follower (1/2) removed (board -1 +
cemetery +1, the M7 dimension); un-selected control (6/7, life > 5) untouched and
still on board (M6 routing; select_count=1 hits only the selected target). 8/8
green; engine 0 errors; check_drift clean; ZERO new Engine/shim/manifest work —
the death path inherited M7's death-voice fix; the predicted damage-VFX shadow
never materialized.
Load-bearing (M4/M6 discipline): swapping the selection to the 6/7 -> it survives
at 2 and nobody dies, proving removal is gated on the SELECTED follower's life
reaching <= 0, not on selection (M7's destroy) or a blanket wipe.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
First proof that follower DEATH / board-removal commits in the authoritative
part of PlayCard headless (not the cosmetic post-Process tail). Card 800144120
(cost-0 when_play destroy of a select_count=1 enemy follower) resolves via the
M6 selectedCards path: selected enemy follower removed (board -1 + cemetery +1),
un-selected untouched (routing confirmed load-bearing by swapping the selection).
Shim gap fixed (the predicted M7 cost): SkillProcessor.SelectCardToHaveDestroyVoicePlay's
cosmetic death-voice tail NRE'd on three M1 default!/Null* shadows
(IBattleCardView.VoiceInfo, CardVoiceInfoCache.GetCardVoiceInfoForBattle,
ReadOnlyVoiceInfo.GetDestroyVoice — the last unusable as the interface since
m1_stub_gen dropped its : IReadOnlyVoiceInfo base). Fix = one hand shim
HeadlessVoiceInfo : IReadOnlyVoiceInfo returning the engine's own
VoiceAndWaitTime._nullVoice sentinel, wired into the two generated seams with
// HEADLESS-FIX markers. No Engine/ edit (drift clean).
dotnet test SVSim.BattleEngine.Tests -> 7/7 green; check_drift.py clean; engine 0 Error(s).
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
First card to exercise the selectedCards path of ActionProcessor.PlayCard
(dormant through M2-M5, all of which played selectedCards: null). Spell
800134020 (clan-1 cost-1, when_play damage=5 to a select_count=1 enemy
follower) resolves headless: with two vanilla followers on the enemy board
and one passed as selectedCards, the damage hits ONLY the selected follower
(13->8) and the un-selected one is untouched (7).
New oracle dimension: SELECTION ROUTING via a differential life-delta on two
surviving targets (selected -5, un-selected 0) — reads the authoritative
damage path M3 proved, with no dependence on follower death/board-removal
timing. Load-bearing confirmed (M4 discipline): swapping which follower is
selected makes the damage follow the selection (assertions fail for the right
reason), then reverted to green.
Like M4, a clean milestone: NO new engine/shim work — the selectedCards path
resolved on the existing shim surface. The only authoring was test-side: the
M6 card constants, a shared HeadlessEngineEnv.PutFollowerInPlay primitive
(create via the null-view seam + drive HandCardToField), and the oracle.
Engine still 0 errors; check_drift clean; dotnet test -> 6/6 green.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Card 800134010 (clan-1 cost-1 ungated spell, summon_token=100011020): a when_play
summon places one new neutral 2/2 follower token on the caster board. New oracle
dimension = board-count + token-identity delta from a SKILL-CREATED card. 5/5 green;
engine 0 errors; check_drift clean; zero new Engine copies.
This is the first headless run of the PUBLIC prefab card-creation path
(CardCreatorBase.CreateCard, createNullView:false) — engine-internal card creation
(summon/draw/token) has no null-view path in solo mode, unlike the M2-M4 hand-card
seam. Built that path headless:
- Self-consistent no-op Unity object graph (UnityShim.cs): Component.gameObject/
transform, GameObject.transform, Transform.parent/Find now lazily non-null +
cached; GetComponent routed through the GameObject component model.
- Targeted NGUI material backing-field wiring (UIFont.mMat / UILabel.mMaterial) so
the copied material getters return non-null via their simple branch (blanket/deep
wiring would make them delegate down a re-nulling chain).
- getUIBase_CardManager() default! -> field-wired no-op via new ShimView.Create<T>().
- Test-side seeds: SBattleLoad card templates + 3D scene GameObjects (InitCardTemplates).
Load-bearing proof: swapping to the M3 non-summoning spell fails the board-count
(Expected 2, was 1) + token-not-found assertions; reverted to green.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Fold SetupCardEvent into a shared HeadlessEngineEnv.CreateHeadlessHandCard primitive
(consolidating the duplicated M2/M3 helpers), then add the M4 oracle: card 103111050
(ELF cost-1 1/1, when_play powerup add_offense=1&add_life=1 to target=self). New oracle
dimension = the played card's OWN stat delta (1/1 -> 2/2). Gate play_count>2 seeded via
the public AddCurrentTrunPlayCount; proven load-bearing (without the seed the fanfare
gates out and Atk stays 1). No new shim/data gaps were needed — only harness seeding.
Engine still 0 errors; check_drift clean.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Card 900124030 (ELF cost-3, when_play damage=3 to enemy leader) resolves to
correct authoritative state headless via the IsForecast/IsRecovery +
ActionProcessor.PlayCard path. New oracle dimension (opponent leader-life delta)
passes; 3/3 tests green; engine still 0 errors; check_drift clean.
Four headless gaps, each mechanical (no logic/Unity wall):
- Data seam: InitLeaderLife (SetupInitialGameState->InitializeClassLife subset);
leader BaseMaxLife was 0 => game-over => play silently rejected. M2 missed it
(only asserted leader life unchanged: 0==0).
- Runtime cast: re-attach IClassBattleCardView on the generated
NullClassBattleCardView stub (members already present; base-clause recovery
stripped the decl). Compiled fine -> M1 loop never surfaced it.
- M1 mis-cut: copy NullVfxWithLoading verbatim (its GetInstance() lazy singleton
was stubbed to default!/null). Same pattern as M2 NullCardVfxCreator.
- Card events: CreateHeadlessHandCard now calls SetupCardEvent so a spell's
OnPlay->RemoveSpellCardFromHand / OnFinishWhenPlaySkill->AddSpellCardToCemetery
fire (the bare CreateCardWithoutResources seam skips them).
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
First green: a zero-skill vanilla follower (100011010, neutral 1/2) resolves to
correct authoritative state HEADLESS via IsForecast/IsRecovery + ActionProcessor.
PlayCard (DP4), no Unity runtime. §5 oracle passes (PP-cost; hand->in-play;
atk/health == CardCSVData base; opponent unchanged; no exception). VERDICT: the
port approach is validated through the resolution path, not just M1's compile path.
VanillaFollowerOracleTests.Vanilla_follower_resolves_to_correct_state — GREEN.
HeadlessCardMaster now loads the follower's real id from cards.json.
Resolution-path shim/engine gaps closed (all mechanical no-op fills or data seams,
never a Unity/logic wall):
- M1 mis-cut copies (DP1/DP3 — pure no-op logic wrongly stubbed to null):
Engine/Wizard.Battle.View.Vfx/NullCardVfxCreator.cs (its GetInstance() singleton
was nulled) + its dep NotEmptyNullVfx.cs. Deleted the generated NullCardVfxCreator
stub + its _IfaceImpl block; both manifested, check_drift clean.
- _IfaceImpl explicit-impl shadow: interface-typed view/mgr calls dispatch to the
explicit impls (which returned default!), shadowing public stubs. Fixed
IBattlePlayerView.GetSideLogControl (SkillProcessor side-log tail) to return a
non-null no-op. KEY M3+ learning: fix _IfaceImpl.g.cs for interface-typed NREs.
(GameMgr/component-model/Resources/IClassBattleCardView shim fills + CardIconControl
copy + the SVSim.BattleEngine.Tests project landed in the prior commit 2b50657.)
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
First green of the M2 go/no-go probe: `new SingleBattleMgr(StandardBattleMgr-
ContentsCreator)` now builds the two-player pair fully headless against the shim,
no Unity runtime. Verdict: headless construction is feasible; every blocker was a
mechanical no-op shim fill or data seam, not a Unity/logic wall.
Shim fills (authored):
- GameMgr: lazy non-null DataMgr/PrefabMgr/InputMgr/SoundMgr/BattleControl.
- GameObject: lazy cached component model so GetComponent<T>/AddComponent<T> return
non-null no-op instances for Component-derived T (F1: unguarded view touches).
- Resources.Load(string): cached non-null GameObject so the prefab->Instantiate->
GetComponent chain (UnityEventAgent) yields a real object.
- ClassBattleCardViewBase: re-attach dropped IClassBattleCardView (no-op members);
ClassBattleCardBase.Setup casts the created view to it.
Engine copy (DP1/DP3 mis-cut fix):
- CardIconControl.cs copied verbatim (manifested) + generated null-stub deleted.
SplitAndCompleteIconStr is pure string logic on the resolution path that M1 had
wrongly stubbed as "View" -> null deref in SkillCreator.CreateBuildInfo.
Test harness (SVSim.BattleEngine.Tests, authored fixture):
- HeadlessContentsCreator/HeadlessPhaseCreator: deterministic replica of the solo
practice init (StandardBattleMgrContentsCreator + SingleBattlePhaseCreator) with
no-op recovery/replay managers.
- HeadlessCardMaster: reflects the loader cards.json dump into CardMaster.
- HeadlessMasterData: minimal Data.Master (class-character list, empty collections)
+ Data.Load + player/enemy chara ids.
- ConstructionProbeTests.SingleBattleMgr_constructs_headless — GREEN.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
