Switches verified inactive before first settlement; moving gains hold (near 43.7 -> 55.5%, far holes 1.8 -> 0%, p99
7.21 -> 6.48 ms, worst 12.32 -> 12.27 ms). Arrival median +1.28% (4/7 pairs slower) though the switches did no work
then; defaults remain off (WaitForAllCriticalTiles=1, ApplyBudget=0). Source/: VoxelWorld.cpp, VoxelWorld.h.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_016dhxM43dnWvUu2B5Ph1iG1
With StreamingFastTravelWaitForAllCriticalTiles=0 and ApplyBudget=1 ms (test values; defaults stay 1 and 0), 7 pairs
vs ca0bf76: near 43.3 -> 56.1%, forward 15.5 -> 17.9%, far holes 1.8 -> 0.0%, p99 7.06 -> 6.52 ms, worst 12.2 ->
11.8 ms, every run p99 <= 7.4 and worst <= 13.8 ms; arrival 1.671 -> 1.683 s (slower in 7/7) so not promoted.
Source/: VoxelWorld.cpp, VoxelWorld.h.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_016dhxM43dnWvUu2B5Ph1iG1
7 runs each: L157 near 55.4% but a 16.7 ms worst tick (UpdateChunks/level sort/overlay); L160 tick-safe but lower
all-around near and later arrival; both: forward cone 25.2% and p99 <= 11 ms, but worst tick up and a 14.3 ms run.
Also found: voxel.StreamingIncrementalDesiredSet defaults to 0, so L165/L166 gains need it enabled.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_016dhxM43dnWvUu2B5Ph1iG1
Persistent Core/overlay/candidate sets updated from deltas; same-center priority order merges changed keys (full sort
only on center change). Set and order equality audits 28/28 + 28/28, injected mismatch fails. 30 m/s UE, 7 pairs vs
9c70053: p99 10.52 -> 7.42 ms (all pairs), worst tick 12.98 -> 11.65 ms (6/7), near 42.0 -> 43.7% and forward cone
14.8 -> 16.1% (all pairs), far holes 3.20 -> 2.74% median (4/7), arrival flat. Source/: VoxelWorld.cpp, VoxelWorld.h.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_016dhxM43dnWvUu2B5Ph1iG1
BuildDelta emits entering/leaving slabs per LOD (incl. coarser-ring coverage changes); cached Core key set updated
from the delta at one-tile crossings, full rebuild for larger jumps or LOD>=4 changes. Equality audit vs full rebuild
(45/45 crossings, 20/20 sampled; injected mismatch asserts) + 630 automation cases. 30 m/s UE, 5 pairs:
BuildDesiredTiles p95 7.80 -> 4.08 ms, p99 11.46 -> 7.97 ms, worst tick 13.73 -> 12.11 ms (all pairs), near and
forward readiness and arrival better in all pairs; far holes better in 4/5 (pair 5 +0.29 pp). Geometry unchanged.
Source/: VoxelClipmapDesiredTiles.h, VoxelWorld.cpp, VoxelWorld.h, Tests/VoxelForgeClipmapNoOverlapTest.cpp.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_016dhxM43dnWvUu2B5Ph1iG1
30 m/s UE, 5 pairs: forward-cone readiness 14.2 -> 18.2% (+3.75 pp), all-around 39.5 -> 36.6%; p99 <= 11 ms in all
candidate runs, but worst tick, far holes and arrival regressed in some pairs. Rebuilds per crossing 7.2 -> 4.4; per
rebuild cost unchanged (not yet incremental). Source/: VoxelWorld.cpp, VoxelWorld.h (defaults unchanged).
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_016dhxM43dnWvUu2B5Ph1iG1
30 m/s UE, 5 pairs: near coverage 36.1 -> 48.6% but 2/5 runs had p99 15.8/18.2 ms (worst 19.9 ms), so controls stay
off. Measured: only ~23% of desired LOD0-2 tiles lie in the 55 deg forward cone; forward readiness is 12.1% (base) /
16.9%, vs 50.9/68.5% behind. Slow frames are dominated by desired-set rebuild/UpdateChunks, not apply. Source/:
VoxelWorld.cpp (default behaviour unchanged).
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_016dhxM43dnWvUu2B5Ph1iG1
Root cause of liminal seed 5 (measured): passage carve makes the bore air (+3.52), then the liminal route fill stamp
re-solidifies it (-0.16) where the deck crosses the tube. Three fixes cleared the bore but left 28-29 road stations
unsupported; none kept. New gate: zero solid samples inside any passage interior, with no-passage negative controls.
Owner v227 hashes and fast tier pass.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_016dhxM43dnWvUu2B5Ph1iG1
Layers may publish certified arrival/departure pads; the passage planner binds endpoints and tangents to them;
no-pad layouts serialize unchanged (owner canonical hashes match v227). Pads removed the endpoint grade failures;
seeds 0-4 passed at a 45 m margin, but seed 5 still has 289 blocked passage-capsule samples ~29 m before the
departure pad, where the composed field stays solid (-0.162) inside the tube over native ground (-3.6). Ripple
audit now keeps true route distance across masked gaps (with a synthetic control).
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_016dhxM43dnWvUu2B5Ph1iG1
Plans and draws flat field+monolith, repeating hills, road over cloud floor and pillar grid in a 3-strata fixture,
binding passage/0-1 and 1-2 as fixed endpoints. Composed W1 probes: 3/4 seeds fail one passage interface, 4/4 fail the
liminal walk audit near the mouths (support itself passes). Stopped per AGENTS.md. Same mouth-ownership failure
class as FI (L133/L140) and VS seed 2 (L139).
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_016dhxM43dnWvUu2B5Ph1iG1
Road: one swept-deck stamp on a shared cubic-Hermite centerline (stamp vocabulary V5), 376.8 m, max 8.5 deg, walk
audit clean, short-road control fails as expected. Pillars: regular fillPillar stamps get deterministic 24-sided
capped-cylinder proxies once the LOD step exceeds their radius or half-height (applies to all worlds). Owner hashes
unchanged (v227); five-fixture Clang/GCC determinism; fast tier green.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_016dhxM43dnWvUu2B5Ph1iG1
Captures stream LOD0-8 (hills repeat at 80 m to the horizon). Road: cubic swept solid deck, 376.8 m, 8 m wide, max
12.06 deg, 17 supports, cloud floor ~45 m below; walk audit 0 blocked/unsupported, 300 m negative control fails as
expected. 15x15 identical pillar grid at 76 m. Owner canonical hashes unchanged; fast tier, Clang/GCC green.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_016dhxM43dnWvUu2B5Ph1iG1
Standalone only: fixtures, liminal_road_override (fillDais), --audit-liminal-road-walk, capture-only distance fog.
Road 65 m x 6 m, 22 m up, capsule walk 127/127. Owner hashes unchanged (v227), fast tier, Clang/GCC green.
Hills repeat every 80 m but the capture reaches 56 m, so only one rise shows. Also R48 (who owns the tunnel mouth):
recommends a bounded joint plan of passage ports and layer landing candidates.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_016dhxM43dnWvUu2B5Ph1iG1
--density-dispatch-report prints each strate's bound path and tile shares. Owner world: SW native heightfield;
FI, TN, VS on compiled-analytic legacy fallback (FI lacks an imported FloatingIslandPlan; TN/VS have legacy recipes).
A native FI candidate broke the composed owner walk/shell gates, so the owner stays on its path. Output unchanged.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_016dhxM43dnWvUu2B5Ph1iG1
One current-state page (shipping behaviour, references with sources, gaps, verification, closed lines), linked from
CLAUDE.md. Removes the L86/L88 passage/boundary ablation switches and the unused legacy route-grade audit; inventory
of every kept switch in the report. Output identical: 24/24 + 18/18 hashes, fast tier, walk/seam audits, UE host
build; timing unchanged (-0.2%, CI spans 0). No Source/ changes.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_016dhxM43dnWvUu2B5Ph1iG1
Reach = 0.75 shell + 0.21 relief + 0.50/6 smooth-max support + 0.01 = 1.0533 m, used by proof, feature, fallback,
off-grid bounds and the sweep cutoff. Removes the dead 4 m term (v225 added metres to core-voxel coordinates).
Captures and mouth close-ups byte-identical; proof audit 0 misses over 25,728 shell/crest samples; floor 7.884 mm;
seams, walkability, fast tier, Clang/GCC W1/W8/W12 green; timing equal within noise.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_016dhxM43dnWvUu2B5Ph1iG1
Pair overlap comes from per-segment bounds expanded by radius + 4.335 m route reach; two exact filters cut
floor-profile calls 20-23% but near-ready stayed +3..+11% vs L109 (within ~2% of L125).
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_016dhxM43dnWvUu2B5Ph1iG1
Conservative route slabs with an outside sentinel, tight expanded segment bounds; DensityGrid exact fallbacks
479k -> 0 per run. Near-ready 2-14% faster than L119, still +4.7..+10.2% vs L109; whole view within -0.5..+8%.
Same look (floor offset 7.884 mm, captures checked), seam audit bit-identical, fast tier + walk audits,
Clang/GCC W1/W8/W12 determinism green. BVH and 25 cm tolerance trials rejected. Not for merge yet.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_016dhxM43dnWvUu2B5Ph1iG1
Shape gate is real (no LOD0-2 tile before shape ready, ~180-226 ms) but the shape pool is already ~8/8 busy;
independent 27-tile pages net at most ~62 ms (VS negative). Near-ready today: SW 0.69-0.71 s, VS 0.84-0.99,
TN 1.24-1.37, FI 1.33-1.44 s. No runtime change.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_016dhxM43dnWvUu2B5Ph1iG1
The 18 ms gate is one game-thread tick, not compile cost (0.25 ms). MPSC handoff keeps ownership but still
polls per tick; velocity look-ahead didn't help (near -0.8 pt, p99 11.77 > 11.00). Both kept as a patch.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_016dhxM43dnWvUu2B5Ph1iG1
98% of idle-with-holes windows wait on the expanded critical set; region compile runs 0.26 ms but its
gate stays pending 18 ms p50. Broad early admission: near +15.4 pt but p99 13.4 ms (over the VR 11 ms
frame); near-only: +9.9 pt but far holes +1.5 pt. Trace instrumentation kept as a patch.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_016dhxM43dnWvUu2B5Ph1iG1
After the normal visible-apply count is reached and <50% of the 2 ms budget is spent, apply one horizon
result (off during startup boost; voxel.StreamingFarApplySupplement=0 disables). 30 m/s UE, 5 pairs:
far holes 4.38% -> 2.70%, far finish-to-apply p50 1.105 -> 0.114 s, worst tick 14.74 ms (<15.43),
near coverage +0.3 pt, arrival whole-view -313 ms. Geometry unchanged (24/24 hashes), fast tier green.
Adds trace-only RMC proxy publication timing.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_016dhxM43dnWvUu2B5Ph1iG1
Far holes coexist with idle workers: far tiles wait median 1.12 s / p95 3.21 s from worker completion
to apply while generating in p95 10 ms, so the game-thread result queue and admission are the limit.
Priority split failed; 10 slots helped but broke the worst-tick bar. LOD2 still waits for far parents.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_016dhxM43dnWvUu2B5Ph1iG1
Chord error 2.5 -> 10 cm, shell work skipped outside the 0-0.75 m band. Sweep pairs -39%, prepare -60%,
near-ready 3-8% faster than L115 but still +13..+22% vs L109. Floor offset 7.88 mm; captures match L115
by eye; fast tier, walk audits, Clang/GCC W1/W8/W12 determinism green. Not for merge yet.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_016dhxM43dnWvUu2B5Ph1iG1
Tagged the passage bucket: per-tile prepareSweptTubeBase dominates (0.47-0.92 s summed per view),
smooth-min second. Segment lower-bound pruning (-17% pairs) and deferred fallback floor work stayed
hash-exact but moved near-ready <3%. Still +17..+26% vs L109.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_016dhxM43dnWvUu2B5Ph1iG1
Near +18..+32% is ~95% in passage-feature tiles (12-17% of tiles): sweep pairs 2.3-2.6x, exact
fallbacks +71..+101%, segment visits 3-4x, plus an untagged cache/prepare/smooth-min bucket
(+0.47 s/view). Skips, triangle counts (-0.1..-1.4%), skirts, normals, far shells ruled out.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_016dhxM43dnWvUu2B5Ph1iG1
Fine pairs 48.9M -> 38.8M (-21%) with L115-identical hashes, but no readiness change (still
+15..+30% near-ready vs L109). Pair evaluation is not where the regression lives. Kept as patch.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_016dhxM43dnWvUu2B5Ph1iG1
Floor height from the planner profile h(s) by horizontal projection + arc-length interpolation;
adjacent fine segments share endpoint values. Floor zero-crossing offsets max 7.8 mm (L114: up
to 73 mm); rock-floor close-ups show no ribs. The 2 L114 shell misses were oracle false positives
(points inside an overlapping bore); the oracle now checks every overlapping span; no gate relaxed.
Fast tier PASS; Clang/GCC W1/W8/W12 deterministic. Speed regression grows: near-ready +17..+31% vs
L109 (fine-segment/sample pairs 49.0M vs 19.8M). Not merged.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_016dhxM43dnWvUu2B5Ph1iG1
Floor ribs are present in L109/L112/L113 rock tunnels too: the floor height comes from the
nearest segment, so it jumps (up to 5.9-7.3 cm) where the nearest segment switches. v222 reduces
them; faint ribs remain. Quiet-machine timing (load 0.64): near-ready 9-17% slower than L109
(FI W12 0.95 -> 1.10 s); candidate totals ~5x L109. Fast tier FAIL: 2 SW wall-shell probes of
31,381 read host air. Walk/grade audits pass; Clang/GCC deterministic; UE builds.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_016dhxM43dnWvUu2B5Ph1iG1
Open-air walls: 0.75 m rounded shell, relief fades over the top 0.25 m - the needle spikes are
gone. Every owner plan passes the 15 deg forward-grade check (max ~10.7 deg) and all walk audits.
Floor bands: median 1.25 m interval, ~0.21 m drop on a ~10.4 deg floor = one 0.25 m voxel row
projected along the grade (not control points, not slab thickness) - not yet fixed. Exact
segment-to-tile filter did not reduce candidates; removed. Timings this round are unreliable
(a gitea process used ~43% CPU throughout). Fast tier PASS; Clang/GCC deterministic; UE builds.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_016dhxM43dnWvUu2B5Ph1iG1
Dense centripetal spline in XY, height piecewise-linear on the planner's certified profile,
radius/floor depth blended; open air = floor slab + 1 m side walls, no roof; audits sample the
generated centreline. All walk audits pass (grades <= 15 deg, headroom, shell, mouths, ripple);
Clang/GCC W1/W8/W12 deterministic; fast tier PASS; Linux UE builds. Chevron rings gone.
Open issues: jagged spikes along the half-pipe wall tops, visible step lines on the trough
floor; near-ready +7..+10% vs L109 (candidate list p95 4 -> 22). Not merged.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_016dhxM43dnWvUu2B5Ph1iG1
Route resampled to ~1 m (310 -> 2,790 segments); open air writes a 0.75 m floor slab plus 1.0 m
side walls, no roof. The 3D spline overshoots between planned route points, raising floor grades
above 15 deg; a grade-limited variant left 2 unsupported samples on passage 0-1, and the shell/
ripple audits still model the old chord route. Stopped after two designs. Not merged.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_016dhxM43dnWvUu2B5Ph1iG1
Owner design: flat world-horizontal floor (0.50 m corner blend), 0.75 m solid shell written only
where the host field is air (interiors union first and beat every shell), 5 m cubic mouth ramps.
Removed route support restoration, deferred support, support columns, mouth-support restoration
and their capsule checks (-5,129 / +1,641 lines). Near-ready @W12: FI 1.39 -> 0.86, TN 1.29 ->
0.90, VS 0.85 -> 0.74, SW 0.70 -> 0.76 s (more shell/skirt surface). All walk audits pass;
Clang/GCC W1/W8/W12 deterministic; fast tier PASS; Linux UE builds; 6 UE arrival traces pass.
OPEN ISSUE: open-air shell (e.g. passage 0-1, 1037-1489 m route span) shows a regular stepped
banding pattern - not merged into assembler/world until fixed/approved.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_016dhxM43dnWvUu2B5Ph1iG1
export_world.py exports the owner four-strate world (LOD0-2 around each arrival + one FI
passage view) through the headless mesh exporter and converts to GLB (16-45 MB, 0.6-1.7M
triangles each); GLBs are git-ignored and regenerated by the script. Menu collapses to a
button (collapsed by default on phones), smaller phone controls, loading progress and errors.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_016dhxM43dnWvUu2B5Ph1iG1
Policy 3 movement traces (12 s move + 5 s dwell, 3 runs each). Unreal 30 m/s: near LOD0-2
ready while moving 31.6 -> 44.2% (LOD0 15.4 -> 34.7%), world tick p99/max 23.1/28.3 ->
11.3/13.6 ms, dwell completion 3.55 -> 3.22 s; far holes 2.5 -> 4.3%. Walking unchanged (~99%).
Stale work is small (1.9% of moving LOD0 tiles finish out of range); the remaining limit is
sustained generation throughput (8 task slots, p95 worker 105 ms). Near:far stays 4:1.
World unchanged (pinned hashes pass). Fast tier PASS; Linux UE builds.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_016dhxM43dnWvUu2B5Ph1iG1
Fr0zka
