Squashes 146 commits from battle-engine-extraction. Net: 2,045 files changed,
+11,896 / -158,687 lines. Ships engine passes 4-7 (dead-code cull, view-layer
stub, receive-path shrink) plus the Phase-5 AsyncLocal ambient deletion that
turns concurrent battles into a type-system property rather than a scope
contract.
## What landed
**Passes 4-7 (chunks 1-34):** Extended the Phase-4 const-false collapse into a
cascading cull across the skill graph, view layer, and receive-path periphery.
Six mode flags (IsWatchBattle/IsReplayBattle/IsAdmin/IsAdminWatch/IsPuzzleQuest/
IsAINetwork) became `const false`, every guarded block deleted. Field*.cs
subclass ctors + BackGroundBase + ObjectChecker culled to no-ops. Mulligan
family reworked to take a mgr param through IMulliganMgr.InitMulligan.
Emotion/Recovery/Resource clusters null-stubbed. Prediction/OperationSimulator/
skill filters converted from static ambient reads to per-mgr reads via
SkillPrm.ownerCard.SelfBattlePlayer.BattleMgr / ins.BattleMgr / this.BattleMgr.
**Phase-5 ambient rip (chunks 35-47):** Deleted BattleAmbient / BattleAmbient-
Context / TestBattleScope in full. Every per-battle mutable slot now lives on
the mgr instance itself:
mgr.InstanceIsForecast / InstanceIsRandomDraw / InstanceRecoveryInfo /
InstanceViewerId / InstanceNetworkAgent / GameMgr
BattleManagerBase.GetIns() returns null unconditionally; the residual static
flags + 3 façades (Certification.ViewerId, Data.BattleRecoveryInfo,
ToolboxGame.RealTimeNetworkAgent) are null-tolerant defaults kept for the
handful of engine-internal readers that still reference their types. Zero
BattleAmbient references anywhere in engine + node + tests.
Added pre-seeded GameMgr ctor overload threaded through the mgr chain
(BattleManagerBase → SingleBattleMgr / NetworkBattleManagerBase → NetworkStandard-
BattleMgr → HeadlessBattleMgr / HeadlessNetworkBattleMgr). Fixtures build a
GameMgr, seed it via HeadlessEngineEnv.SeedCharaIds/SeedNetUser, and pass it
to the mgr's ctor — no ambient reach.
Node side (SVSim.BattleNode/SessionBattleEngine): _ctx replaced with a plain
GameMgr field; 34 `using var _ambient = BattleAmbient.Enter(_ctx)` scope wraps
ripped from every accessor and mutator; EngineGlobalInit.WirePerSessionGameMgr
takes GameMgr as a param and runs from SessionBattleEngine.SetupInternal
BEFORE mgr construction.
Test side: TestBattleScope deleted; 18 fixture [SetUp]s migrated to
`HeadlessEngineEnv.EnsureProcessGlobals()`; MultiInstanceEngineTests rewritten
around per-mgr construction (GetIns() → null is the pinned invariant).
## Regression fixes
- **chunk-48** (MulliganCtrl): chunk-35's `= null` stubs on card lookups broke
the live receive-driven Deal path (BattlePlayerBase.DrawCard NRE'd downstream
of NetworkPlayerMulliganCtrl.StartMulliganVfx). Restored the three lookups
via `_battlePlayer.BattleMgr.GetBattleCardIdx`. Engine tests were satisfied
by the WireMulliganPhase seam; unit tests exposed the live-path gap.
## Ship state
- SVSim.BattleEngine.Tests: 56/56 pass, 2 skip
- SVSim.UnitTests: 1554/1554 pass (was 1523/31-fail before chunk 48)
- Solution build: 0 source warnings (40 pre-existing NU1902 MessagePack CVEs
in SVSim.EmulatedEntrypoint, unrelated)
- Sequential PVP smoke: verified live (two back-to-back battles, no regression
on cleanup/spinup)
- Concurrent PVP smoke: verified live
Adds tools/engine-port/ClosureAnalyzer/ — the Roslyn transitive-type-closure
analyzer needed to make future cascade cleanup safe (per feedback memory
"Engine cleanup needs closure tool" from the 2026-06-28 pass-3 failure).
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Reviewer noted the factory may be invoked more than once under contention.
Document the analysis inline so a future reader doesn't have to redo it:
the discarded instance's mutations land on private fields of a soon-unreachable
object, and the only shared sentinel (_noopViewMaterial) is read-only.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
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>
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>
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>
Hygiene fixup for the IsForecast/IsRandomDraw ambient conversion in 3b5f2e1.
The manifest sha was stale (pointed at the pre-ambient RNG-virtual-patched
contents) and the change had no companion .patch artifact alongside
BattleManagerBase.rng-virtual.patch. Follow established convention.
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>
Drive ATTACK frames through the headless receive conductor and assert on engine
board state (node-native harness). Two cases: follower -> enemy leader (leader
life drops by atk, attacker spent) and a lethal follower-vs-follower trade (both
removed). ATTACK opcode confirmed = 10 (NetworkBattleDefine.PlayActionType).
Headless view-untangle (no Engine logic edits; drift clean):
- IBattlePlayerView.AttackSelectControl -> non-null HeadlessAttackSelectControl
(no-op RegisterAttackPair/ResetCardAfterAttack); IsCardTranslatable left to base.
- IBattleCardView.CardInfo -> backing card via BuildInfo (so IsCardTranslatable
reads authentic IsClass); class/null view ctors now chain : base(buildInfo).
- IBattleCardView._inPlayFrameEffect -> non-null no-op control.
- Seed Certification.viewer_id headless so the IsRecovery target parse
(vid != UserViewerID) does not throw inside SavedataManager and silently drop
the parsed targetList.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
- _IfaceImpl.g.cs: extend header to warn about hand-edits; tag all bare
// HEADLESS-FIX lines with their milestone (M13 on GetSideLogControl ×2)
so `grep HEADLESS-FIX` reliably surfaces every block before a regen.
- HeadlessHandViewStub / HeadlessPlayQueueViewStub: narrow from public to
internal sealed — both stubs are consumed only within SVSim.BattleEngine
(via the generated partial impls); no public surface exposes the concrete
type, so internal is correct and aligns with HeadlessIconAnimations.
- SessionBattleEngine.SeedMulliganInfoControl: add one-line comment on the
GetComponent<MulliganInfoControl>() call explaining the shim's lazy
materialisation behaviour (otherwise reads like a guaranteed NRE).
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
