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>
Build a compare-only dispatcher classifier from RenderFrameView and compare complete opaque, alpha, selection, clip, light, texture, transform, and draw payloads against the accepted path. Preserve retail's stable equal-CYpt ordering with explicit draw-local submission ordinals so material-group history cannot affect alpha ties.
Replace scheduler-quantized software sleeps with a reusable Windows high-resolution deadline timer, expose pacing in the frame profiler, and make shutdown wake every persistent mesh worker without losing the shared signal.
Preserve retail alpha order while using a stable radix, skip duplicate deferred-alpha SSBO packing, pack light sets, cache static selection descriptors, and retire historical material groups at the whole-frame boundary. The fixed dense-Caul sample improved from roughly 9-12 ms CPU to 5.3-6.2 ms without reducing visual quality.
Release build succeeds with zero warnings and all 6,300 tests pass with five intentional skips. Three independent retail, architecture, and adversarial reviews are clean; the post-review connected route remains pending because local ACE is offline.
Co-authored-by: OpenAI Codex <codex@openai.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>
Port the retail high/low material-lighting cadence for successful world clicks, keep it instance-scoped in the modern renderer, and restore authored lighting after 0.8 seconds. Correct the selection oracle and pin timing plus per-frame buffer lifecycle with tests.
Co-authored-by: Codex <codex@openai.com>
Root cause (measured, not guessed): the OOM was a MANAGED Large-Object-Heap leak
(~5 GB in 6 min of roaming; working set 1.6 GB -> 7.6 GB), dominated by ~3.8 GB of
live System.Single[]. A heap-retention analysis of a captured gcdump traced 241 of
242 giant float arrays to a single List<float>: WbDrawDispatcher.InstanceGroup.Opacities.
#188 (2026-07-09, fading doors) added Opacities as a 5th per-instance parallel list
alongside Matrices/Slots/LightSets/IndoorFlags, but left it out of the per-frame
clear loop (WbDrawDispatcher.cs:959). So Opacities.Add(1.0f) fired once per drawn
instance per frame and the list never reset — growing forever; List<float> capacity
doubling produced the observed ~128 MB / ~512 MB power-of-two arrays and OOM after
~50 min. Hit both UI builds because it's core render, and started only recently
because Opacities did not exist before #188.
Fix: extract the per-frame reset into InstanceGroup.ClearPerInstanceData() (promoted
InstanceGroup private->internal) that clears ALL FIVE parallel lists in one place, so
a future 6th list can't silently drift out of the frame lifecycle again. Line 959 now
calls it. TDD: InstanceGroupClearTests asserts every parallel list resets (RED with
the old 4-list clear, GREEN after adding Opacities.Clear()).
Static-audit note: this is exactly why we measured before fixing — the audit ranked
GlobalMeshBuffer (GPU) #1 and the entity-dict leak #3; both were wrong about the
dominant cause. The dotnet-counters (managed vs native) + gcdump retention graph
named the real one-line culprit.
Secondary (NOT this commit): _groups dictionary is never pruned (bounded by distinct
group keys, empty groups are cheap) — filed as a follow-up, not the multi-GB driver.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
