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>
Final three clusters:
- RoomParamKey: copy Wizard.RoomMatch/RoomParamKey.cs verbatim (UriNames/WatchUriNames
static dicts keyed by PlayerController.ROOM_URI + PlayerControllerForWatching.
SEND_PARAMETER — both now real enums).
- CardChooseTask: copy the TwoPick/CardChooseTask.cs (TaskManager `using`s .Arena.TwoPick,
not .Competition — copy_loop had only landed the Competition twin).
- SetCardNumLabel CS1739: decompiler param-name artifact — the local fn's 3rd param was
recovered as `flag` but call sites pass it named `isRed:`. First DP5 tracked patch:
Engine/UICardList.cs edited (flag->isRed, zero logic change), recorded in
Patches/ + manifest patched=1 (drift-clean).
M1 exit criteria met: `dotnet build SVSim.BattleEngine` = 0 errors, no Unity ref in csproj,
check_drift clean. Session 7: 198 -> 0 across waves 7a-7k.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
- Material.SetVector(int nameID, Vector4); Plane(Vector3, float d) ctor; Socket.On
(SocketIOEventTypes, callback) overload.
- Global GetValueOrDefault(this IDictionary<,>) extension — the BCL form only binds to
IReadOnlyDictionary, so the IDictionary call mis-resolved to JsonDataExtension.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
- VFX: SequentialVfxPlayer.GetAllVfxAsList, ParallelVfxPlayer.GetVfxList,
VfxMgr.CheckAndAddEffectVfxList; point our own ShowBattleUIImmediatelyVfx stub at
NullVfx.GetInstance() (it called a non-existent NullVfx.Create).
- Static factories: SkillTargetSelectTouchProcessor.Create (10-arg),
DialogReportToManagement.Create(long).
- Re-add StartSkillSelectVfx.OnStart event (m1_stub_gen drops `event` decls — the
recurring session-6 gap; generator event-capture fix still pending).
- Stop the BattleCardView/GameObjMgr ctor cascade: parameterless ctors on the no-op
BattleCardView and GameObjMgr hand shims so non-chaining subclass/field stubs satisfy
their implicit base() call.
- Copy Cute/ListExtensions.cs (FisherYatesShuffle extension) verbatim into Engine.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
- Add the (JsonData) ctor to the LoginBonus data hand stubs (Continuous/Normal/
Special/FreeCardPackBox) and StoryRecoveryData (LitJson.JsonData is copied).
- Full-surface the two nested View types that only the parent's empty stub covered:
BattleCardView.BuildInfo (14-arg ctor) and DestroyVfx.FileNamePair (ctors +
ObjectFileName/SeFileName); add BattleCardView(BuildInfo) to the hand shim.
- Regenerate Field/Spell/UnitBattleCardView: stale stubs whose ctors had dropped the
decomp `: base(buildInfo)` chain, exposed (CS7036) once BattleCardView lost its
implicit default ctor.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
- Generate both SelectionProcessing Parameter classes namespace-aware (full-surface
captures the 8-arg Main / 6-arg BattleResult ctors); drop the empty hand stubs.
- Add the missing decomp ctors to the 5 empty Touch-processor hand stubs
(SetCard/EvolutionSimple/Emotion/ClassBuff/DetailPanel) — compile-only ballast,
empty bodies.
- Regenerate FusionWaitProcessor.g.cs: it was a stale stub whose ctor had dropped
its decomp `: base(...)` initializer; harmless while SetCardProcessor had an
implicit default ctor, exposed (CS7036) once the parameterized ctor landed.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
The Story chapter-selection processing subsystem is duplicated across two
namespaces (…SelectionProcessing.Main and .BattleResult), each with its own
ProcessingBase : IProcessing + Parameter. m1_genstub keyed output by bare type
name, so only ONE ProcessingBase.g.cs was emitted (BattleResult), and
m1_baseclauses cross-qualified the Main leaves to BattleResult.ProcessingBase —
making it impossible to give IProcessing its real members (Execute(Main.Parameter)
≠ inherited Execute(BattleResult.Parameter) → CS0535).
Now both ProcessingBase variants are generated via the namespace-aware tooling
(<Type>__<Namespace>.g.cs), baseclauses resolves each leaf to its same-namespace
ProcessingBase, and both IProcessing interfaces carry NextProcessing + Execute.
8 IProcessing CS1061 cleared, no CS0535 introduced.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
base-clause recovery strips interfaces (to dodge CS0535), but copied code converts
the stubs to those interfaces -> ~120 CS0266/CS1503. Two mechanisms:
- _IfaceImpl.g.cs: explicit no-op impls of the FULL (copied) interfaces, layered
onto each hierarchy base (BattleCardView/CardVfxCreatorBase/BattlePlayerView/
BattleEnemyView/ClassInfomationUIBase + NullCardVfxCreator). Explicit form never
collides with existing members; leaves inherit. Walks base-interface chains
(IPlayerView : IBattlePlayerView) and emits events.
- _InterfaceReattach.g.cs: plain ': IFoo' for the empty stub interfaces
(IProcessing, IReplayRecordManager).
- ClassBattleCardViewBase/NullBattleCardView: restore dropped BattleCardView base
so they inherit its IBattleCardView impl.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
BestHTTP.SocketIO: Socket.On/Off/Emit + SocketIOCallback delegate, SocketManager ctors/
State/Socket/indexer/Open/Close/SettingRealtimeNetworkAgent, SocketOptions. Spine: Skeleton
(Data/Skin/Scale/FindBone/SetSkin/Update), Bone (WorldX/Y/RotationX), SkeletonMecanim
(MonoBehaviour + skeleton). All minimal hand shims (no full-surface -> no SDK closure pull);
node-socket path is Phase-2, off the battle path.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Make the minimal hand shims partial + generate full member surface for the manager/
task/controller god-objects (LoadingViewManager/DeckUpdateTask/MyPageTask/ReplayController/
PlayerControllerForWatching/WatchDataHandler/EvolutionTouchProcessor/StoryChapterSelection
Utility/NonDialogPopup). NonDialogPopup given MonoBehaviour base + hand Close() removed
(superseded by full surface). LoadTask dup deleted (already copied verbatim). RoomMatch
watch/replay closure types stubbed. Copied 8 more closure files.
CS0246-in-generated-signature masking note: 4 such errors were hiding ~1582 — generated
CS0246 masks as hard as header CS0246; the real frontier is 1586 (CS7036 base-ctor +
member-level), 0 structural.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Resolves the 268-error header frontier: settings Item base, ErrorDialog.Data,
RoomConnectController nested types, Unity asset/light/collider types, CriWare/
CodeStage/Spine SDK surface, and copies INetworkLogger + SingletonMonoBehaviour
verbatim. Per F3 this unmasks the type bodies (~26.5k member-level errors now
visible) -- the real M1 bulk, attacked in following waves.
