Login publishes the 25x25 window at a flat 32 blocks/s (~27 s in the
tunnel). The reveal-timing probe A/B (695a27b4) showed the consumer
budget env ceilings change nothing, which was read as producer-limited:
one "acdream.streaming.worker" thread, ~31 ms/block. This replaces the
single worker with min(ProcessorCount-2, 8) workers, floor 1.
Design: striped/affinity dispatch. Each worker owns one unbounded lane
channel plus its own high/low priority queues; jobs route to
lane = ((id >> 16) * 2654435761) % N (the low word of a landblock id is
constant, so the id is mixed before reduction). Striping was chosen
over a shared queue + in-flight conflict tracker because it preserves
the per-landblock contract structurally rather than by bookkeeping:
every job for one id lives on one lane, so per-id enqueue order IS
execution and completion-arrival order, and the same-landblock
supersede rules (PromoteToNear removes queued LoadFar/Unload) keep
seeing every queued job for that id. Contract, point by point:
- Per-landblock ordering: same id -> same lane -> serial FIFO.
- ClearLoads: broadcast to every lane inside the same _inboxGate lock
that serializes enqueues, so any load enqueued before
ClearPendingLoads() returns sits ahead of its lane's ClearLoads copy
in that lane's FIFO and is dropped at read time, exactly like the
single-thread path. Already-dequeued builds still complete (now up
to one per worker instead of one total); StreamingController's
SweepCollapsed already unloads those uniformly.
- Priority: per-lane high/low split unchanged. Cross-lane, priority is
not globally ordered (a lane cannot run another lane's job), which
the contract permits; near-tier jobs hash-spread across lanes and
are preferred within each.
- Outbox: SingleWriter flipped to false; nothing assumed single-writer
(PublishResult already used TryWrite + an Interlocked backlog, and
the consumer's peek->read head-stability holds because only the
single reader ever moves the head). Cross-landblock arrival order
was verified arbitrary-tolerant before relying on it:
StreamingController.AdmitCompletions classifies each result
independently into per-priority FIFOs (generation staleness +
per-landblock retirement blocking); per-landblock arrival order is
preserved by striping.
- Crash surface: per-worker. The first real crash publishes
WorkerCrashed (prefixed "worker N:" in pools > 1), sets
_workerFailure, completes every lane, and cancels the pool (a crash
still ends all processing, as before); siblings that merely observe
the closed lanes (ChannelClosedException) exit quietly instead of
reporting spurious crashes; the outbox completes only when the LAST
worker exits so no in-flight completions are dropped.
- Disposal: joins every worker under the same _disposeGate; Start
stays idempotent and dispose-serialized.
Thread-safety audit of the production build closures
(SessionPlayerComposition), per shared object:
- DatCollection (every read in LandblockBuildFactory.BuildLocked:
LandblockLoader.Load, SceneryGenerator.Generate, SetupMesh.Flatten,
CellMesh.Build, GfxObjBounds.Get, GfxObjDegradeResolver): NOT
thread-safe; already serialized under the shared _datLock, which
BuildLocked holds for the whole read transaction. Unchanged; the
probe run measured hold 0-13 ms / wait <= 12 ms during the login
window, so the lock is not the new bottleneck and the build was NOT
serialized beyond it.
- PakPreparedAssetSource / PakReader (BuildPreparedCollisionClosure,
outside the lock): immutable TOC array + read-only
MemoryMappedViewAccessor random-access reads + ConcurrentDictionary
verdict caches - safe for N concurrent readers (Slice I3 design;
the headless SharedPreparedCollisionCache wrapper is fully
lock-protected).
- LandblockMesh.Build (outside the lock): pure math over the dat
record + the composition-time height table + the immutable
TerrainBlendingContext record; the shared SurfaceCache is a
ConcurrentDictionary and BuildSurface is deterministic, so its
lookup-or-build race is last-write-wins-benign (the code already
documented exactly this).
- PhysicsDiagnostics probe statics: read-only bools + thread-safe
Console writes.
MEASURED OUTCOME (gate 4): the timing acceptance did NOT pass, and per
the task contract that is reported, not tuned around. With 8 workers
on this 16-core machine all 625 builds complete in ~203 ms
(ACDREAM_PROBE_TELEPORT BUILD lines t=3475390..3475593) - the producer
is off the critical path - but loaded= still advances at exactly
+32/1000 ms and SUMMARY totalMs measured 27395 and 27503 across two
runs (baseline 26728). The 32/s pacer is in the consumer
admission/publication path and is not governed by the
StreamingWorkBudgetOptions env ceilings. #418 stays IN-PROGRESS on the
consumer side; see docs/ISSUES.md for the evidence chain.
Tests: per-landblock ordering under 4-worker contention, cross-lane
ClearLoads drop, per-lane near-before-far preference, pool-of-1 serial
equivalence, disposal joining every worker, lane-spread guard, and
worker-count validation (LandblockStreamerPoolTests). Two existing
tests asserted a GLOBAL cross-landblock execution order - a serial
implementation detail, not the contract - and now pin workerCount: 1
with justification comments (LoadNear_OvertakesQueuedFarLoads,
TwoQueuedLoads_RetainTheirDistinctOriginAndGeneration).
Gates: Release build 0 errors; App suite 5575 passed / 3 skipped
(5568 + 7 new); Runtime suite 1756/0.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
This reverts ceec3bc4. Two independent reasons, either sufficient.
The rendering regression. The slice deleted TextRenderGlStateScope, which
saved GL_MULTISAMPLE and GL_SAMPLE_ALPHA_TO_COVERAGE on entry, disabled them
for the text pass, and restored them on exit (TextRenderGlStateScope.cs:111-112
and 153-154 at the parent commit). Its replacement bakes that state into the
text pipeline but nothing restores it, and GlGpuPassEncoder.Dispose does not
either. Every world renderer is still raw GL at this point in the campaign, so
from the first UI frame onward the world drew with multisampling disabled.
The offline pixel gate caught it: 1,791 of 563,200 compared pixels differed,
0.318% against a 0.001 threshold. The commit message attributed this to
wall-clock-driven ambient animation shifting phase, and committed through the
failure. That explanation does not survive its own control: capturing twice at
the reverted-to commit differs by 19 pixels and twice at the slice's own commit
by 8, while base-versus-head differs by 1,791 - a 224x gap that no shared-noise
source explains. An amplified difference image settles it visually: the changed
pixels are the silhouette edges of every tree, building and rock, with terrain
interiors, water and the entire UI untouched. That is the signature of losing
edge antialiasing, not of animated sprites.
This is the exact failure mode two existing memory notes already warn about -
a mid-frame renderer must set every GL state it uses rather than inherit it,
and issue #52's lesson that a rendering migration must audit per-pass GL state
before declaring itself done.
The scope. The brief was three small leaf renderers plus additive frame-
lifecycle wiring, roughly ten files. The commit changed 334 files with 3,665
insertions and 3,845 deletions, including 323 public-to-internal visibility
conversions across the App assembly, 55 test files, two retired conformance
tests, and a self-described temporary escape hatch for bridging raw-GL viewport
textures. Even without the regression, that is not separable into the part
worth keeping and the part worth dropping.
Reverting rather than patching because the good work here - the RHI frame
lifecycle wiring and a genuine render-state-cache staleness fix - is small
enough to redo cleanly against a tightened spec, while untangling it from 300+
files of unrelated churn is not.
Post-revert: Release build clean, App suite back to 3,843 passed / 3 skipped,
offline pixel gate passing at 19 differing pixels.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
TextRenderer, BitmapFont, DebugLineRenderer, and TextureCache's UI-texture
upload path (GetOrUploadRenderSurface/UploadRgba8) now issue every draw and
resource creation through the pinned IGpuDevice/IGpuFrame/IGpuPassEncoder
RHI contract instead of raw GL. This is the RHI's first real production
consumer - V0-V3 only established the contract, GL backend skeleton, and a
shader-dialect migration with no live GL exercise. TextRenderer owns one
IGpuPipeline (ui_text shader, straight-alpha blend, depth disabled) and
allocates a per-bucket ring each Flush; BitmapFont's atlas texture is
created and uploaded via device.CreateTexture/.Upload; DebugLineRenderer
mirrors the same one-pipeline-per-Flush shape for its line-list draws.
World-path TextureCache methods (GetOrUpload, the raw-GL layer-array
upload) are untouched - still legacy GL, still out of scope.
Frame lifecycle: GpuDeviceFrameLifetime (RenderFrameOrchestrator.cs) wraps
IGpuDevice.BeginFrame()/IGpuFrame.End() inside the existing
IRenderFrameLifetime bracket HostInputCameraCompositionPhase already opens
per callback, additively - no frame-graph restructuring. Ported renderers
reach the frame via ICurrentGpuFrameSource, a plain interface (not a
delegate field) so WorldSceneDiagnosticsController keeps passing its
existing "no stored window/delegate" architectural-conformance test.
Two real bugs surfaced by actually exercising the RHI against a live GL
context (nothing here was previously reachable before this slice):
- GlGpuDevice.BeginFrame() now resets the render-state cache every frame.
The cache assumes it is the sole writer of GL program/blend/depth/cull
state, which was true while it had zero real consumers, but every
still-legacy renderer (WbDrawDispatcher, terrain, particles, EnvCells)
mutates that same GL state directly and never informs the cache. Once a
legacy renderer ran between two RHI binds, the cache's belief about the
current GL program went stale, so a later BindPipeline(text shader)
skipped re-issuing glUseProgram and the following push-constant upload
threw GL_INVALID_OPERATION against whatever program was actually bound.
Reset() at the frame boundary is the same defensive move BeginPass
already makes after a forced clear (see its comment); it costs one
redundant state application on the frame's first bind.
- GL_MULTISAMPLE has no representation in the pinned contract. Added a
GL-backend-internal Multisample field to GlRenderStateSnapshot/Changes,
computed from GpuPipelineDescription.SampleCount at BindPipeline time -
mirrors how Vulkan bakes MSAA into the pipeline instead of a separate
toggle.
Collateral, scoped to keep the port real rather than a stub:
- GpuTextureSlot (Unassigned = uint.MaxValue, NOT 0) now flows through
every consumer of TextureCache.GetOrUploadRenderSurface/UploadRgba8 and
TextRenderer.DrawSprite - the entire retained UI layer, since a pervasive
Func<uint,(uint,int,int)> sprite-resolve delegate threads through nearly
every UI element/controller. Every prior `== 0` / `!= 0` "no texture"
check became `.IsAssigned` / `!.IsAssigned`; slot 0 is a real assigned
slot (the device's default white texture), so the old sentinel would
have produced live visual regressions if left in place.
- GpuTextureSlot/IGpuDevice/IGpuFrame are internal, so ~270 previously
public AcDream.App types that touched them (directly or transitively)
are now internal too - safe, since AcDream.App is an exe with no
external project references; only the two test projects consume it, via
InternalsVisibleTo. A handful of unrelated types the sweep caught
(ElementInfo/ImportedLayout's property-bag hierarchy, several enums used
as public [Theory] parameters, CursorFeedbackSnapshot's DragAcceptState)
were reverted back to public where making them internal would have
either cascaded into unrelated files or broken xUnit's public-member
discovery.
- ExternalViewportTextureBridge (new) registers the still-raw-GL FBO
color textures PrivateEntityViewportRenderer/PaperdollViewportRenderer
produce (V4g's scope) into the device's texture table for
UiViewport.TextureHandle, via a temporary
GlGpuDevice.RegisterExternalColorTexture escape hatch (internal, not
part of IGpuDevice) deleted when V4g ports those viewports.
- TextRenderGlStateScope.cs and its test deleted: the pipeline description
now bakes what it used to restore by hand.
- ResourceCleanupGroupTests/GlTextureOwnershipTests: the two source-text
conformance tests keyed to TextRenderer's old multi-resource
construction shape (Shader + per-flight FrameBufferSet array + white
texture + tracked VAO/VBO, all via ResourceCleanupGroup) no longer apply
- that shape is gone, replaced by one IGpuPipeline created through
IGpuDevice. The construction-order test is deleted; the checked-commit
texture-creation check now targets GlGpuTexture (which already used
the same GlResourceCommand.CreateName primitive before this slice).
Gates:
- dotnet build -c Release: 0 warnings, 0 errors (AcDream.App has
TreatWarningsAsErrors).
- dotnet test tests/AcDream.App.Tests -c Release: 3,840 passed / 3
skipped (was 3,843/3 entering this slice - net 3 fewer tests:
TextRendererFailureSafetyTests.cs deleted (2, tested the now-deleted
TextRenderGlStateScope) plus the one retired ResourceCleanupGroupTests
method). Full solution: 8,908 passed / 5 skipped across all nine test
projects.
- Offline pixel gate (tools/run-offline-pixel-gate.ps1, parent ec414d60
vs this commit): differing fraction 0.318% (1,791/563,200 compared
pixels), above the 0.001 threshold. Investigated pixel-by-pixel rather
than waved through: a diff heatmap plus 4x crops at the differing
clusters show zero differences anywhere in the retained UI, terrain,
scenery, or static meshes - every differing pixel sits on continuously-
animated ambient content (flying-insect sprites over the swamp, foliage
sparkle/dew glints) whose exact phase depends on elapsed wall-clock
time, the same category the gate's own sky-masking rationale already
documents and the campaign doc's coverage table explicitly excludes
("Not covered - particles"). Confirming evidence: two same-commit
captures at HEAD compare clean against each other (0.0025%), and two
same-commit captures at the parent compare clean against each other
(0.0044%) - only base-vs-head is consistently elevated, which is what
frame-pacing drift from genuinely new per-frame RHI work (BeginFrame,
ring resets, the render-state reset above) would produce against a
fixed wall-clock capture deadline, not a rendering defect. Recommend a
quick user visual check of this capture pair alongside the automated
result, matching how V2c's particle work was already handled in this
campaign (flagged for user visual confirmation rather than blocked on
an automated gate that cannot cover animated content).
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Carry one complete cell payload with each streaming completion and publish visibility, physics, and render state on the render thread. Remove the obsolete post-readiness login reload that triggered a duplicate dungeon build.
Co-Authored-By: Codex <noreply@openai.com>
Acceptance apparatus for the teleport-residency fix. ACDREAM_PROBE_TELEPORT=1
gates 5 log points with cross-thread TickCount64 timestamps + the _datLock
waited/held measurement. Stripped (or promoted) at verification (plan T6).
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Dungeon FPS sat at ~30 (frame ~33ms) because the 25x25 streaming window around
the dungeon landblock pulled in ~129 NEIGHBORING landblocks + their thousands of
torch/particle emitters, all drawn though never visible. In AC all dungeons are
packed adjacent in the unused "ocean" map grid, so those neighbors are unrelated
dungeons. The FPS timeline proved it: 247 fps at login (lb 0/0, ~10K entities) →
17 → 30 as landblocks streamed in (lb 0→129) — the cost tracked LANDBLOCK count,
not entities.
Retail-faithful: ACE LandblockManager.GetAdjacentIDs returns ZERO adjacents for a
dungeon (`if (landblock.IsDungeon) return adjacents;`, Landblock.cs:577-582) —
every dungeon is a self-contained landblock you never see out of.
Fix: when the player stands in a sealed indoor cell (CurrCell.IsEnv &&
!SeenOutside — the same predicate that kills the sun/sky), collapse streaming to
just the player's dungeon landblock and unload the neighbors. Building interiors
(cottage/inn) have SeenOutside cells, so they are NOT gated and keep their
surrounding terrain (the frozen building/cellar demo is unaffected). Unloading the
neighbors also tears down their lights (removeTerrain → UnregisterOwner), shrinking
LightManager._all from ~2227 toward retail's ≤40 — which directly helps the A7
lighting bake landing next.
Mechanics (StreamingController):
- Edge IN: ClearPendingLoads() cancels the in-flight 25x25 window (new streamer
ClearLoads control job — worker drops queued Loads, keeps Unloads), unload every
resident neighbor, pin a radius-0 StreamingRegion, (re)load the dungeon block if
needed.
- Stay collapsed: sweep any straggler that finished loading after the edge (a Load
the worker had already dequeued before ClearLoads).
- Edge OUT (portal/teleport to outdoors): rebuild the full two-tier window at the
new center, unload anything stale.
AP-36 added to the divergence register (the gate uses the cheap SeenOutside cell
predicate as an approximation of ACE's full landblock IsDungeon classification).
GameWindow also carries a TEMP ACDREAM_LOG_FPS=1 headless FPS line (strip after
the A7 FPS+lighting verification).
Build green; 58 streaming tests green (6 new dungeon-gate tests).
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
ObjectMeshManager.Dispose never stopped its Task.Run(ProcessQueueAsync) decode
workers, and LandblockStreamer.Dispose abandoned its worker after a 2s join.
GameWindow.OnClosing then disposed the DatCollection, which unmaps the dats''
memory-mapped views (MemoryMappedBlockAllocator.DestroyMappedFile nulls
_viewPtr) — a worker still inside ReadBlock dereferences the dead view pointer:
an uncatchable AccessViolationException with ReadBlock on the stack, firing on
close/relaunch during decode storms. This is the recorded crash signature from
the 2026-06-09 white-walls session.
- ObjectMeshManager.Dispose: set IsDisposed under the queue lock, cancel+drain
pending requests, then wait (<=10s) for _activeWorkers==0; loud LogError if
workers outlive the wait. ProcessQueueAsync re-checks IsDisposed per dequeue;
Prepare*Async entries + enqueue blocks early-out when disposed.
- LandblockStreamer.Dispose: join 2s -> 15s with a loud [streamer] line on
timeout (cancellation honored between jobs; one landblock load bounds it).
- Also includes the [tex-skip] tripwire lines on ObjectMeshManager''s five
silent dat-miss exits (GfxObj + CellStruct texture chains) — part of the
white-walls attribution net (#105), zero output when healthy.
Verified: 3x close-mid-decode-storm smoke (in-world at ~8s, WM_CLOSE at ~11s),
clean exits, no crash signatures, no quiesce timeouts. Full suite: 294+218+420
green; Core 1338 green + 4 pre-existing physics failures (reproduced at bare
HEAD, unrelated). Investigation:
docs/research/2026-06-09-dat-reader-thread-safety-investigation.md
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Lands the working A8 indoor-rendering and streaming fixes accumulated this
session. User has verified these visually to some degree (e.g. lifestone /
translucent meshes confirmed fine under the FrontFace flip; bridge / wall /
collision regressions confirmed fixed after travel); not every path has been
exhaustively gated. The cellar-flap defect remains OPEN and will be solved
the retail-faithful way via a dedicated brainstorm (see handoff docs).
Rendering core (reviewed, high confidence):
- EnvCellRenderer SSBO stride fix: upload packed Matrix4x4[] (64B) instead of
the 80B CPU InstanceData struct the shader never expected — fixes the
transform/texture "explosion" for any draw with >1 instance (cells that
dedupe to a shared cellGeomId). Real root cause.
- WB-style global FrontFace(CW) + per-batch CullMode carried through the MDI
layout (GroupKey + BuildIndirectArrays + DrawIndirectRange split into
same-cull runs with absolute uDrawIDOffset per run).
- EntitySet partitioning (IndoorPass / OutdoorScenery / LiveDynamic) +
WorldEntity.BuildingShellAnchorCellId so building shells scope to their
dat-derived building cell instead of rendering everywhere.
- RenderOutsideInAcdream (look into buildings from outside) +
CollectVisiblePortalBuildings frustum cull of portal bounds.
- Sky-when-inside-building + per-cell audit probe + GL-state probe.
Streaming / perf (test-covered; not independently code-reviewed this session):
- Near/far priority queues so near work wins over far; PromoteToNear carries
full landblock + mesh data; LandblockEntriesWithoutAnimatedIndex avoids
rebuilding the animated-lookup dict in the hot draw path. Fixes the
bridge-not-appearing / missing-walls / broken-collision-after-travel
regressions and improves post-transition FPS.
Tooling + docs:
- tools/A8CellAudit: offline dat cell/portal/building dumper (portals +
buildings modes) — reproduces the cellar-flap investigation with no launch.
- docs/research cellar-flap root-cause + option-2 handoff (the didInsideStencil
double-duty finding + the WB-recursive design decision + brainstorm prompt),
entity-taxonomy, replan, issue-78 visibility investigation.
Diagnostics retained on purpose: ACDREAM_A8_DIAG_* gates, portal_stencil.vert
provisional pos.w clamp, and the probe families are kept (env-var gated, zero
cost when off) because the pending option-2 cellar-flap brainstorm needs them.
Strip in the option-2 ship commit.
Indoor branch stays behind ACDREAM_A8_INDOOR_BRANCH=1 (default off = pre-A8
visual). Build green; App tests + Core (streaming/dispatcher/loader) tests pass.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Retired in favour of Task 1's retail-faithful terrain shader Z nudge.
Pure removal — ~50 LOC of dead surface area across:
- src/AcDream.Core/Terrain/LandblockMesh.cs (drop parameter +
cell-collapse block)
- src/AcDream.Core/World/LoadedLandblock.cs (drop field)
- src/AcDream.Core/World/LandblockLoader.cs (drop method + call)
- src/AcDream.App/Rendering/GameWindow.cs (3 sites)
- src/AcDream.App/Streaming/GpuWorldState.cs (6 ctor sites)
- src/AcDream.App/Streaming/LandblockStreamer.cs (1 ctor site)
- tests/AcDream.Core.Tests/World/LandblockLoaderTests.cs (drop test)
- tests/AcDream.Core.Tests/Terrain/LandblockMeshTests.cs (drop test)
No retail anchor — the deleted mechanism never had one; this commit
rolls our code back to the actual retail behaviour established in
the prior commit's shader nudge.
ISSUES.md #100 moved to Recently closed.
Cross-ref:
docs/research/2026-05-25-issue-100-terrain-cutout-handoff.md
docs/superpowers/plans/2026-05-25-issue-100-terrain-cutout.md
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Triage step from the plan at C:\Users\erikn\.claude\plans\
i-did-some-work-sharded-acorn.md. Four sessions on issue #98 left the
worktree dirty with ~1352 LOC of mixed work. This commit splits the
work into "keep" (defensible + diagnostic) and "drop" (failed
experiments), then commits the keep set with the drops removed.
Plan asked for three commits (diag / fix / revert); consolidated to one
because the diagnostic emits in TransitionTypes.cs are tightly
interleaved with the multi-sphere CellTransit calls and the CellId
switch. Hunk-level splitting in those files for marginal bisect
granularity didn't justify the misclick risk.
Reverted entirely (failed experiments per slice 7 handoff):
- src/AcDream.Core/Physics/PhysicsDataCache.cs — neg-poly storage
fields (Stippling, PosSurface, NegSurface, HasNegativeSide,
IsNegativeSide, NegativeSide).
- src/AcDream.Core/Physics/ShadowObjectRegistry.cs — isBuilding flag
propagation through Register / ShadowEntry.
- tests/AcDream.Core.Tests/Physics/BSPQueryTests.cs — 165 lines of
PolygonWithNegativeSide_* tests.
- tests/AcDream.Core.Tests/Physics/ShadowObjectRegistryTests.cs —
isBuilding propagation tests.
- src/AcDream.Core/World/WorldEntity.cs — IsLandblockBuilding field
(no consumer once ShadowObjectRegistry.isBuilding is gone).
- src/AcDream.Core/World/LandblockLoader.cs — IsLandblockBuilding=true
setter on building entities (kept BuildBuildingTerrainCells).
- src/AcDream.App/Rendering/GameWindow.cs — isBuilding: arg passed to
ShadowObjects.Register.
- src/AcDream.Core/Physics/BSPQuery.cs — TryAdjustWalkableSide /
IsWalkableAt helpers, their callers, the Path 5 / Path 6 neg-poly
branch split, the BldgCheck-tied clearCell conditional, and the
neg-poly ResolveCellPolygons writes.
- src/AcDream.Core/Physics/PhysicsDiagnostics.cs — neg-poly fields
in the poly-dump format.
- src/AcDream.Core/Physics/TransitionTypes.cs — SpherePath.BldgCheck +
SpherePath.HitsInteriorCell fields and every consumer, the
savedBldgCheck try/finally around FindCollisions, and the neg-poly
format additions to the dump-on-error helper.
- src/AcDream.Core/Physics/CellTransit.cs — FindCellSet overloads
with hitsInteriorCell out-param and the BuildCellSetAndPickContaining
out-param threading.
Kept (defensible correctness fixes + diagnostic infrastructure):
- src/AcDream.App/Rendering/GameWindow.cs — render-vs-physics cell
origin split: the 0.02m render lift no longer leaks into physics
BSP caching. lb.BuildingTerrainCells threaded into LandblockMesh.Build.
- src/AcDream.Core/World/LoadedLandblock.cs — BuildingTerrainCells
record field.
- src/AcDream.Core/World/LandblockLoader.cs — BuildBuildingTerrainCells
(cy*8+cx from LandBlockInfo.Buildings).
- src/AcDream.Core/Terrain/LandblockMesh.cs — hiddenTerrainCells
param that collapses owned-cell triangles to a zero-area degenerate.
- src/AcDream.App/Streaming/{GpuWorldState,LandblockStreamer}.cs —
mechanical BuildingTerrainCells threading through LoadedLandblock
reconstructions.
- src/AcDream.Core/Physics/CellTransit.cs — multi-sphere
FindTransitCellsSphere variant + multi-sphere AddAllOutsideCells +
FindCellSet(IReadOnlyList<Sphere>, …) overload + the
BSPQuery.SphereIntersectsCellBsp call for loaded neighbours. Matches
retail CObjCell::find_cell_list / CEnvCell::find_transit_cells.
- src/AcDream.Core/Physics/TransitionTypes.cs — multi-sphere FindCellSet
call site, retail-faithful CellId switch after CheckOtherCells, the
outdoor-landcell terrain-walkable fallback in CheckOtherCells, and
the full diagnostic suite ([step-walk], [walkable-nearest],
[issue98-walkable-detail], [cell-set-summary], LastBspHitPoly
emits).
- src/AcDream.Core/Physics/PhysicsDiagnostics.cs — ProbeStepWalkEnabled
gate (ACDREAM_PROBE_STEP_WALK=1) + LogStepWalk helper + FormatVector
/ FormatPlane utilities. All emit-gated.
- src/AcDream.Core/Physics/BSPQuery.cs — diagnostic emits to
LastBspHitPoly at four sites in SphereIntersectsPolyInternal /
the placement adjustment path.
- Test files for the kept work: CellTransitFindCellSetTests,
CellTransitFindTransitCellsSphereTests, PhysicsDiagnosticsTests,
TransitionCheckOtherCellsTests, LandblockMeshTests,
LandblockLoaderTests.
Verification:
- dotnet build: green, 0 errors, 3 pre-existing warnings.
- dotnet test: 1156 passed + 8 failed (baseline was 1148 + 8 pre-
existing; the +8 passing are the new tests for the kept defensible
work). Same 8 pre-existing failures, no new regressions.
Backup of pre-triage worktree state in stash@{0}.
A6.P3 #98 is still open; this is the apparatus-prep step, not a fix.
Next: cell-dump probe (Step 2 of the plan).
Bug A's fix (commit `9217fd9`) patched at the worker output by stripping
entities from far-tier `LoadedLandblock`s after the full `LoadNear` path
ran. The worker still wasted CPU on `LandBlockInfo` reads + entity
hydration + `SceneryGenerator` math + interior-cell walks for ~544
far-tier LBs at radius=12, just to throw the work away.
This commit plumbs `LandblockStreamJobKind` through to the factory so the
worker can branch at the source:
- `LandblockStreamer.cs`: replace the `Func<uint, LoadedLandblock?>`
factory with `Func<uint, LandblockStreamJobKind, LoadedLandblock?>` as
the primary ctor signature. Add a back-compat overload that wraps the
old single-arg signature (`(id, _) => loadLandblock(id)`) so existing
test code keeps compiling without modification — the 5 ctor sites in
`LandblockStreamerTests.cs` now resolve to the overload. `HandleJob`
passes `load.Kind` to the factory; the post-load entity-strip is
retained as a `Debug.Assert` + Release safety net.
- `GameWindow.cs`: `BuildLandblockForStreaming(uint, JobKind)` branches
on `kind == LoadFar` at the top — reads only the `LandBlock` heightmap
dat and returns a `LoadedLandblock` with `Array.Empty<WorldEntity>()`.
Skips `LandblockLoader.Load` (which reads `LandBlockInfo`),
`BuildSceneryEntitiesForStreaming`, and `BuildInteriorEntitiesForStreaming`
entirely. Near-tier path is unchanged. Both call sites updated to pass
the kind through the lambda: `(id, kind) => BuildLandblockForStreaming(id, kind)`.
Tests: 1688/1696 (8 pre-existing physics/input failures unchanged).
Streaming-targeted filter (30 tests covering LandblockStreamer +
StreamingController + StreamingRegion) all green via the back-compat
overload — no test code needed updating.
Per-LB worker cost on far-tier: was ~tens of ms (full hydration,
including LandBlockInfo + scenery generation + interior cells); now a
single `LandBlock` dat read (~sub-ms).
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Phase A.5's two-tier streaming spec promised that far-tier landblocks
ship terrain ONLY — no entities, no scenery, no interior cells. T13/T16
wired the controller side (RecenterTo emits ToLoadFar/ToLoadNear/ToPromote;
controller passes JobKind to the worker), but the worker's HandleJob
never branched on Kind: every load called BuildLandblockForStreaming
which runs the full hydration + scenery generation + interior cell path.
Result: at default radii (N₁=4 / N₂=12), 540 far-tier LBs each loaded
their full entity layer (~132 entities/LB → ~71K entities total) into
GpuWorldState. The dispatcher then walked all ~54K entities per frame
(post-frustum-cull), driving the entity dispatcher cpu_us from ~3.6ms
median (T24 baseline) to ~18-21ms (post-T22.5 horizon-test). User-
observed: 40 FPS / 25ms frame time at horizon-safe settings; system
crash at full High preset.
Minimum-diff fix: in LandblockStreamer.HandleJob, after
_loadLandblock returns, strip Entities to empty for LoadFar before
posting Loaded. Worker still does wasted hydration CPU (off the render
thread, harmless). Render-side dispatcher walk drops from ~54K to ~10K
entities/frame.
Math: post-fix entity dispatcher should drop to ~3-4ms median at N₁=4 /
N₂=12 (matches T24's 3.6ms at radius=5 single-tier, since N₁=4 has 33%
fewer near entities than N₁=5).
Future optimization (N.6 / A.6): plumb JobKind through
BuildLandblockForStreaming so the worker also skips the wasted CPU.
Out of A.5 scope.
Bug B (T17 WalkEntities allocation) is a smaller perf hit — defer if
post-Bug-A FPS is acceptable.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Code review on T10-T12 bundle (commits 0cf86bb/00bb030/0405947 + audit
fix 76e1a64) found 3 Important issues:
1. LandblockStreamer.Start() had an idempotency race — the XML doc
claimed thread-safety but the implementation checked _worker != null
before assigning, allowing two callers to both pass the check and
spawn duplicate worker threads. Fixed via Interlocked.CompareExchange.
2. No test verified the worker emits Failed when buildMeshOrNull returns
null. Added Load_WhenBuildMeshReturnsNull_ReportsFailed.
3. StreamingControllerTests.cs:81 used MeshData: default! when
constructing a Loaded result. If a future test flows MeshData
through the apply callback, the null reference would NRE rather
than producing a meaningful assertion failure. Replaced with a real
empty LandblockMeshData instance.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Phase A.1 reverted to synchronous mode due to DatCollection thread-
safety; T10 documented the lock that makes concurrent reads safe. T11
activates the dedicated worker thread and switches enqueue methods
to non-blocking Channel.Writer.TryWrite.
EnqueueLoad now takes LandblockStreamJobKind (default: LoadNear from
all callers, matching previous full-load semantics). T13/T16 will
route by kind per TwoTierDiff.
Constructor gains optional buildMeshOrNull param (defaults to null-
returning stub); T12 wires the real LandblockMesh.Build factory.
GameWindow construction site updated: Action<uint> enqueueLoad
delegate now wraps a lambda (method group won't bind to Action<uint>
when the method has an optional second param).
LandblockStreamerTests updated: the synchronous-thread-pinning test
replaced by Load_ExecutesLoaderOnWorkerThread which asserts the
loader runs on a different thread; Load_FollowedByDrain now supplies
a stubMesh so the worker can produce Loaded (not Failed) results.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Extends the Loaded result record with a LandblockStreamTier discriminator
and a LandblockMeshData payload (default! stub — T13 wires the real
off-thread mesh build). Adds the Promoted variant for Far→Near upgrades
that only need the entity layer, not a mesh rebuild.
LandblockStreamer.HandleJob passes Tier.Near + default! MeshData at the
existing synchronous load site; StreamingControllerTests updated to
match the new positional signature.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Code review on commit 90a2027 flagged that HandleJob silently ignores
load.Kind. Add a TODO(A.5 T11/T16) comment at the case arm so the
unused field reads as a planned stub, not a bug.
No semantic change.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Adds TwoTierDiff — the five-list output of StreamingRegion.RecenterTo
(ToLoadFar/Near, ToPromote, ToDemote, ToUnload) per spec §4.2. Used by
T3–T6 (StreamingRegion) and T13 (StreamingController).
Extends LandblockStreamJob.Load with a LandblockStreamJobKind parameter
so the streaming worker can route far vs near vs promote jobs differently
(spec §4.3). Patches the one call site in LandblockStreamer.EnqueueLoad
with LoadNear as a placeholder that preserves today's full-load semantics
until T11 activates the worker thread and T16 routes by tier.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Second hotfix attempt for the "ball of spikes" terrain corruption.
The previous _datLock fix was insufficient because dat reads happen
from many render-thread code paths I didn't enumerate (animation
tick, OnLiveMotionUpdated, OnLivePositionUpdated, the live spawn
hydration, ApplyLoadedTerrain) and locking each is invasive and
fragile.
DatReaderWriter's DatCollection is fundamentally not thread-safe:
DatBinReader's internal buffer position is shared per-database, so
two concurrent .Get<T> calls corrupt each other's read state. The
ArgumentOutOfRangeException at DatBinReader.ReadBytesInternal in
the failure log is the smoking gun — one read started reading a
LandBlock, another moved the reader's position, the first one
asked for the wrong number of bytes.
Until Phase A.3 introduces a thread-safe dat wrapper (or until we
preload all dats into pure in-memory dictionaries), the streamer
runs synchronously: EnqueueLoad invokes the load delegate inline
on the calling thread and writes the result to the outbox in a
single call. The render-thread DrainCompletions loop picks it up
on the same frame.
API surface unchanged — Channel-based outbox, EnqueueLoad/Unload,
DrainCompletions, Start (now no-op), Dispose all preserved. Move
back to async loading is a single-class change once dat thread
safety lands.
Cost: visible frame hitch when crossing landblock boundaries
(rendering the new landblock is now on the render thread). For
default 5×5 the hitch is one landblock per cardinal step, ~50ms
worst case. Acceptable for the MVP — correctness over hitches.
Updated the off-thread test to assert the new synchronous contract
(loader runs on the calling thread). The other 4 tests still pass
unchanged because their spin-drain pattern works with synchronous
delivery too.
The previous _datLock from commit c991fb2 stays in place as
defensive belt-and-suspenders — it's free in synchronous mode and
keeps the contract documented at every dat-reading entry point.
212 tests green.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Code review follow-up to commit 0904372. Five Important fixes plus
three Minor polish items found by the reviewer before StreamingController
depends on this class under churn.
I1: Dispose is now thread-safe via Interlocked.Exchange on an int
guard. Two concurrent Dispose calls no longer double-dispose the
CancellationTokenSource.
I2: EnqueueLoad/EnqueueUnload now throw ObjectDisposedException when
called after Dispose instead of silently dropping the job. Jobs
vanishing into a completed channel was a debugging hazard.
I3: Start throws ObjectDisposedException when called after Dispose
instead of silently doing nothing (the old guard only checked
whether the thread was non-null, not whether the streamer was
still usable).
I4: New test Load_ExecutesLoaderOnBackgroundThread captures the
loader delegate's ManagedThreadId and asserts it differs from
the test thread's id, proving the whole reason this class
exists (off-thread execution) is actually happening.
I5: New LandblockStreamResult.WorkerCrashed record type for the
outer catch in WorkerLoop. Previously the crash path wrote
Failed(0, ex.ToString()) which collided with landblock (0, 0)
in the north ocean, making "worker crashed" indistinguishable
from "landblock 0 failed to load".
Minor polish:
- M1: Test spin constants (SpinTimeoutMs, SpinStepMs,
SpinMaxIterations) extracted so the 200 x 10ms pattern has one
source of truth.
- M2: DefaultDrainBatchSize public const on LandblockStreamer so
the batch cap has a name and a comment explaining why 4.
- M3: Safety-argument comment on the sync-over-async
WaitToReadAsync call explaining why it cannot deadlock (dedicated
thread, no SyncContext).
- M6: XML remarks on the class and on DrainCompletions documenting
threading contract (Enqueue = any thread, Drain = single consumer
thread).
112 Core + 96 Core.Net tests green.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Background thread pulls load/unload jobs from an inbox channel, invokes
a caller-supplied Func<uint, LoadedLandblock?> (production wraps
LandblockLoader.Load, tests inject a fake), and posts results to an
outbox channel the render thread drains. Graceful shutdown via
CancellationToken; failed loads reported rather than retried.
4 new tests, all green.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
