The rest of V4t. The composite, particle and shared-atlas texture paths now
hand out the device's GpuTextureSlot instead of a raw 64-bit
ARB_bindless_texture handle, and GroupKey, CachedBatch and ObjectRenderBatch
carry that slot. WbDrawDispatcher, EnvCellRenderer and ParticleRenderer retire
their interim GlBindlessHandleTable instances and share the device's one
table, exactly as V4t-1 did for terrain. Nothing about world submission
changes otherwise: these three renderers are still raw GL, still bind binding
9 themselves, and still draw the same geometry in the same order.
**What produces a slot now.** CompositeTextureArrayCache's GL backend interns
each array's handle when it makes it resident and retires the entry when it
makes it non-resident, so the pair is created and destroyed together and the
cache above it never learns a device exists — the fake backend its tests use
mints a stand-in slot. TextureCache.AcquireParticleTexture does the same for
the one-layer particle arrays it owns, including on its rollback path.
ObjectMeshManager registers each shared atlas's wrap/clamp handles at batch
upload; registration is idempotent by handle, so the many batches sharing an
atlas share its entry.
**Slot release is stricter than what it replaces, not looser.** The interim
tables never released anything — the class comment said so — and they grew
without bound. The device's table has a fixed 16,384-slot capacity, so an
unreleased entry is now a leak with an end. Every producer therefore retires
its entry: the composite backend at MakeNonResident, the particle backend at
MakeNonResident, and ObjectMeshManager when a retiring atlas's PHYSICAL
retirement completes — the point at which its handles are already non-resident
and its texture already deleted. That last one needs the handles snapshotted
at eviction, because ManagedGLTextureArray.Dispose zeroes its own copies as
its first act. Teardown deliberately does not release: the device is being torn
down alongside its callers, so there is nothing left to recycle a slot into,
and deferring work through a possibly-disposed retirement queue would turn a
clean shutdown into a throw.
**The default value became load-bearing, and that is the one real hazard here.**
BindlessTextureLocation could say "not resolved" with handle 0, because no
texture has handle 0. A slot index has no spare value — default(GpuTextureSlot)
is real slot 0 — so a positional record would have turned every
budget-rejected or still-uploading composite into a silent read of whichever
texture registered first. That is the magenta-placeholder failure shape one
layer down. The type is now a struct storing the slot one-based, so default IS
Unresolved, with a test pinning both halves: default is unresolved, and a
location naming slot 0 is resolved and distinguishable from it. Elsewhere the
sentinel is already exact — GpuTextureSlot.Unassigned is 0xFFFFFFFF, which is
common.glsl's ACDREAM_TEXTURE_NONE — so the classify path's "no texture yet"
test and the particle billboard's untextured branch are unchanged in meaning.
**GroupKey ordering is preserved because the key never ordered anything.**
Handle→slot is a bijection (the device interns one slot per resident handle),
so the same (entity, batch) pairs bucket together as before. The key reaches
equality, hashing and the scene-digest fingerprints — never a comparator:
opaque and translucent groups sort by cull mode then camera distance, the
delayed-alpha path by viewer distance then submission ordinal, and group
enumeration follows the persistent dictionary's insertion order, which a
changed hash does not disturb. The digests hash the slot index where they
hashed the handle; both sides of the render-shadow comparison compute them the
same way, so the value changing is invisible to it. Read
CompareOpaqueSubmissionOrder, CompareTransparentSubmissionOrder and
AlphaFingerprintComparer before doubting this — sort-order drift is a
pixel-visible regression class this project has hit, and it is why the check
was made before the retype rather than after.
**One visibility change, forced rather than chosen.** BindlessTextureLocation
was public and now holds an internal contract type, so it is internal;
ObjectRenderBatch.TextureSlot is internal on an otherwise public class for the
same reason. Nothing outside this assembly and its InternalsVisibleTo test
assemblies named either.
**SkyRenderer keeps its interim table**, and the report should say why: the
sky's textures are minted by SkyRenderer itself from TextureCache's raw GL
texture names, which this slice does not retype, so it would be the one
consumer registering handles it produced — a different shape from the world
stack. The offline gate also masks the sky band, so the one automated
instrument here cannot see a sky regression. V4f owns that renderer.
**Gates.** GL offline pixel gate vs cb2a70b8, measured twice: 31 and 22
differing pixels of 563,200 (5.50e-05, 3.91e-05). The first is above the
plan's documented 15-23 px band, so a control was measured rather than
assumed: two same-commit captures at this tree differ by 19 px, and — the
decisive number — a capture at V4t-1 and a capture at this commit differ by
9 px, fewer than the same-commit control. Maximum channel delta is 41-52 in
every pair including the controls, i.e. the differing pixels are drawn from
one flickering population, not from moved geometry. tools/run-repeat-connected-gate.ps1
-Runs 3: 3/3 RENDERED on both the desktop witness and the client capture. One
Vulkan composition-host run with VK_LAYER_KHRONOS_validation proven inserted
by the loader: zero errors, zero warnings, converged ownership ledger. App
tests 4,077 / 3 skips and the complete Release suite 9,140 / 5 — both the
4,075 and 9,138 baselines plus the two tests added here.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Queue translucent world GfxObj batches and scene particles in one stable far-to-near stream using transformed DAT sort centers, then drain it at retail's landscape and final-world boundaries. Preserve authored blend, cull, lighting, opacity, and adjacent-only batching so particles behind lifestones are composited through the crystal instead of overpainting it.
Release build succeeds and all 5,914 tests pass with five intentional skips.
Co-authored-by: OpenAI Codex <codex@openai.com>
Preserve canonical live-object ownership across Hidden transitions and remote teleport placement so effects, collision, streaming, and targeting remain synchronized.
User reported: the drudge statue on top of the Foundry (a multi-part
live-spawned entity with AnimPartChange + texChanges) renders only
PARTIALLY — some parts visible, some missing.
Root cause: the dispatcher's slow path skips a MeshRef when
_meshAdapter.TryGetRenderData returns null (mesh still async-decoding
via ObjectMeshManager.PrepareMeshDataAsync). The classified-batches
collector accumulates only the MeshRefs that DID resolve. At entity
boundary, the cache populates with the PARTIAL set. Frame-2 cache hits
serve that partial entry forever — even after the missing mesh loads,
the cache continues to skip those parts because classification never
reruns for cached entities.
Fix: track currentEntityIncomplete during the foreach. Set it true on
any null renderData. At entity boundary (and at end-of-loop), if the
flag is set, DROP the accumulated populate scratch instead of writing
it to the cache. The slow path retries on the next frame; once all
meshes have loaded, the populate fires correctly with the complete
classification.
Adds a regression test pinning the contract — incomplete entities
produce zero cache entries.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
User confirmed via A/B test (ACDREAM_DISABLE_TIER1_CACHE=1) that the
visual bug — buildings rendering up in the air outside Holtburg — is in
the cache wiring, not elsewhere. The matrix math (restPose * entityWorld
== model) was provably correct, so the bug had to be cache key collision.
Stabs were namespaced in commit 71d0edc, but scenery (0x80LLBB00 +
localIndex) and interior (0x40LLBB00 + localCounter) still have the
same 256-overflow risk. Dense LBs outside Holtburg (forest, urban) push
localIndex past 255, wrapping into the lbY byte and creating cross-LB
collisions.
Fix: change the cache key from uint entityId to (uint, uint) tuple of
(EntityId, LandblockHint). The cache is now correct-by-construction
regardless of any hydration path's Id-generation strategy. Defensive
against future regressions in any ID namespace.
InvalidateEntity becomes a sweep (was O(1)), but it's called rarely
(only on live-entity despawn). InvalidateLandblock was already a sweep.
Updated 14 existing cache tests + 1 dispatcher integration test to thread
landblockHint through TryGet / DebugCrossCheck calls.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Task 10 (commit 0cbef3c) called ApplyCacheHit inside the per-(entity, partIdx)
foreach loop, but cachedEntry.Batches is flat across all MeshRefs of the
entity. For a 3-MeshRef static building on frame 2: 3 tuples times 6 cached
batches per call = 18 instances drawn instead of 6. Severe Z-fighting and
3x perf hit on every multi-part static entity (buildings, statues, multi-
MeshRef NPCs).
This is the symmetric mirror of the Task 9 bug fixed at 00fa8ae. Both
spec section 5.2 and the plan describe the foreach as per-entity, but
_walkScratch has been per-tuple since Task 6. The implementation
faithfully ported the buggy spec.
Fix: track lastHitEntityId; the cache-hit fast path fires only on the
first tuple of each entity, and subsequent tuples skip the iteration
body via continue. Adds a regression test pinning the per-entity
amplification invariant.
Caught by code review (subagent-driven-development) before Phase 3
dispatched. The bug was invisible in the no-multi-frame-test 1702/8
baseline; would have manifested as visible Z-fighting on every multi-
part building on second-and-subsequent frames once Task 13 perf gate
captured live runs.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
WbDrawDispatcher.Draw now branches on cache hit before running classification:
on hit, walks the cached flat batch list and appends RestPose times entityWorld
to the matching groups; on miss, runs today's classification and populates
the cache (Task 9). Animated entities skip the cache entirely.
Adds dispatcher integration tests #11 (static entity populates + reuses)
and #12 (animated bypasses) per spec test plan section 7.2, plus the
multi-MeshRef regression test that would have caught the bug fixed in
commit 00fa8ae (cache populate must flush at entity boundary, not per-tuple).
Phase 2 (dispatcher integration) complete. End-to-end caching now live.
Invalidation hooks (Phase 3) ensure correctness across despawns + LB demotes.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Per Phase A.5 spec §4.6 Change #1: when an LB is invisible AND
animatedEntityIds is non-empty, the inner loop walked every entity
in the LB just to find the few animated ones. At ~10.7K entities
(N1=4) that is wasted iteration cost per frame.
Extracted a pure-CPU internal static WalkEntities helper. When LB
is invisible: iterate animatedEntityIds directly and look each up
in a per-LB AnimatedById dictionary (typically <50 animated vs
~10K total). When LB is visible: walk all entities as before.
GpuWorldState.LandblockEntries now yields an AnimatedById map as a
5th tuple field alongside the AABB tuple. Dictionary is built on
each yield (cheap — ~132 entities/LB max). A caching layer is out
of A.5 scope.
WbDrawDispatcher.Draw signature updated to consume the 5-tuple.
GameWindow.cs call site passes _worldState.LandblockEntries which
now yields the 5-tuple — no change needed there.
8 new tests in WbDrawDispatcherBucketingTests cover T17 Change #1
(invisible LB / animated set / neverCull / null frustum) and
T18 Change #2 guard tests (cached AABB / dirty flag / animated bypass).
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
