acdream/tools
Erik db4426d5ef test(render): Campaign V slice V11 step 0 — the #256/#257 discriminator, both arms
Issues #256 (server-spawned signs and portals go invisible after repeated
portal runs, while staying interactive) and #257 (working set grows to
~1.5 GB over the same session) were observed together in one long live
Vulkan session, and both filings demanded the same thing before V11 deletes
the OpenGL backend: run the churn on GL too. Vulkan-only growth or drift
would mean the new arm's resource lifecycle is broken, and deleting its only
reference implementation while that was true would be wrong even with the
cutover signed.

So the discriminator is built and run first, and it can stop the slice.

tools/run-portal-churn-soak.ps1 generates a route of N cycles over three
portal-bearing stops taken from the two existing connected routes, runs it
once per backend from one binary, and measures three things the existing
instruments do not measure together:

  * working set and private bytes, sampled from the OS every two seconds and
    joined to each checkpoint by timestamp -- the client's own snapshot has
    no view of its own working set, which is exactly #257's quantity;
  * the published-versus-live pair already in the checkpoint JSON, because
    "alive in the object table, gone from the presentation" is #256's whole
    symptom and a drift between those halves at the SAME stop across cycles
    is what would show it;
  * a within-arm capture comparison -- cycle 1 against cycles 10, 20 and 30
    at a pinned viewpoint -- plus a difference map and a row histogram,
    because a number cannot tell an absent object from a walking NPC and the
    map can.

Every teleloc carries the identity quaternion so the heading repeats, and
the four determinism levers the differential gate forces are forced here for
the same reason: an unpinned sun would swamp the signal.

Result at 90 transits per arm, 91 checkpoints, zero errors, graceful exits:
neither arm reproduces either symptom. Working set means agree to 1 MiB
(GL 1864, VK 1863) and warm-half drift is NEGATIVE on both (-48.0, -27.0).
GPU accounting is exactly constant per arm. worldEntities holds 10,382 at
all thirty cycles on both. The difference maps show every building, the
portal, the statue and the treeline still drawn at cycle 30.

That refutes the one outcome that would have blocked V11, and it does not
identify the pre-existing bug -- so both issues stay OPEN with the negative
recorded, and the follow-up named: walked portal transits rather than
/teleloc, which do not take the same path into the transit state machine.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
2026-07-28 23:24:45 +02:00
..
A8CellAudit diag(render/physics): flap root-caused to physics rest µm-jitter; refute prior diagnoses 2026-06-08 09:16:12 +02:00
cdb fix(#186): render side-cull mis-sided thin connectors — use dat PortalSide bit, not AABB centroid 2026-07-08 15:06:33 +02:00
cdb-scripts fix(motion): full queue reset on locomotion-cycle direct transitions 2026-05-03 17:00:55 +02:00
diag-logs diag(motion): instrumentation for remote walk↔run leg-cycle bug (Commit A) 2026-05-03 20:38:47 +02:00
dump-keymap docs(research): canonical retail keymap + dump-keymap tool 2026-04-25 23:01:58 +02:00
ghidra_project feat(core): port decompiled AC client physics — CollisionPrimitives + PhysicsBody 2026-04-12 23:54:51 +02:00
InspectCoatTex docs(issues): #37 — Investigation 2 narrows bug to SubPalette coverage gaps 2026-05-05 14:45:50 +02:00
jsonl test(phys): A6.P5 fixture — 3 ticks from live door-stuck capture 2026-05-25 12:43:51 +02:00
pdb-extract CLAUDE.md: adopt the condensed structure, updated to current truth (-503/+176) 2026-06-12 12:42:27 +02:00
PesChainAudit feat(vfx): decode retail hooks and typed tables 2026-07-14 08:17:44 +02:00
ProjectileVfxAudit fix(vfx): classify hardwareless particle emitters once 2026-07-27 00:03:44 +02:00
RainMeshProbe feat(sky): load Setup-backed (0x020xxx) sky objects via SetupMesh.Flatten 2026-04-27 23:24:09 +02:00
RetailTimeProbe sky(phase-8): retail-faithful night sky + README refresh 2026-04-24 20:34:36 +02:00
ShaderCompiler ci(render): make V9's first CI run green on both operating systems 2026-07-28 21:52:53 +02:00
SkyObjectInspect sky(phase-8): retail-faithful night sky + README refresh 2026-04-24 20:34:36 +02:00
StarsProbe tools(probe): add StarsProbe to dump every SkyObject's geometry + UVs 2026-04-26 22:19:28 +02:00
TextureDump fix(combat): restore retail power bar layers 2026-07-12 21:14:14 +02:00
WeatherEnumerator docs(research): sky/weather investigation handoff + diagnostic tools 2026-04-26 21:40:34 +02:00
WeatherSetupProbe docs(research): sky/weather investigation handoff + diagnostic tools 2026-04-26 21:40:34 +02:00
analyze_resolve_capture.py tools+plan(#182): resolve-capture histogram classifier + verbatim player-physics rebuild plan 2026-07-07 13:37:15 +02:00
compile-shaders.ps1 ci(render): make V9's first CI run green on both operating systems 2026-07-28 21:52:53 +02:00
connected-backend-differential.route.txt fix(diag): Campaign V slice V7 commit 2 - pin the world clock, because the route never did 2026-07-28 19:34:13 +02:00
connected-dense-town.route.txt fix(diag): bind stationary samples to route checkpoints 2026-07-24 12:54:57 +02:00
connected-r6-soak.route.txt fix(diag): bind stationary samples to route checkpoints 2026-07-24 12:54:57 +02:00
connected-world-lifecycle.route.txt test(runtime): add connected lifecycle gate 2026-07-20 22:39:34 +02:00
connected-world-reconnect.route.txt test(runtime): add connected lifecycle gate 2026-07-20 22:39:34 +02:00
decompile_acclient.py research: decompile acclient.exe terrain/physics via Ghidra headless 2026-04-12 23:18:27 +02:00
decompile_full.py research: full acclient.exe decompilation — 22,225 functions, 688K lines 2026-04-12 23:25:51 +02:00
DecompileTargets.java research: decompile acclient.exe terrain/physics via Ghidra headless 2026-04-12 23:18:27 +02:00
ghidra_decompile.py research: decompile acclient.exe terrain/physics via Ghidra headless 2026-04-12 23:18:27 +02:00
run-backend-differential-gate.ps1 fix(diag): Campaign V slice V7 commit 2 - pin the world clock, because the route never did 2026-07-28 19:34:13 +02:00
run-blank-world-ab-probe.ps1 docs(render): the occlusion-query verdict on the V4c blank world, and the options 2026-07-28 01:44:34 +02:00
run-blank-world-surface-probe.ps1 test(render): add blank-world attribution apparatus and a post-world GL sample 2026-07-28 00:08:10 +02:00
run-connected-r6-soak.ps1 feat(streaming): shadow-publish flat collision assets 2026-07-25 16:38:54 +02:00
run-connected-world-lifecycle-gate.ps1 feat(render): Campaign V slice V10 — Vulkan becomes the default backend 2026-07-28 22:32:25 +02:00
run-offline-pixel-gate.ps1 feat(render): Campaign V slice V10 — Vulkan becomes the default backend 2026-07-28 22:32:25 +02:00
run-offline-vulkan-capture.ps1 feat(render): Campaign V slice V6j commit 2 - Dereth draws on Vulkan 2026-07-28 15:46:54 +02:00
run-portal-churn-soak.ps1 test(render): Campaign V slice V11 step 0 — the #256/#257 discriminator, both arms 2026-07-28 23:24:45 +02:00
run-repeat-connected-gate.ps1 feat(render): Campaign V slice V10 — Vulkan becomes the default backend 2026-07-28 22:32:25 +02:00
verify_un2_fmul.py UN-2 RESOLVED: GetMaxSpeed x4 is byte-verified retail; doc-comment was the misread 2026-06-12 13:17:50 +02:00