CellTransit.FindTransitCellsBox ports CEnvCell::find_transit_cells'
part-array overload @0x0052cae0 line-for-line: per-portal x per-part
order, the sphere cheap-reject at F_EPSILON+radius, the box admit whose
"Straddle or crossing-side" rule is exactly retail's `eax != side` under
the PDB Sidedness enum, leads-outside placed AFTER the admit, the
unconditional unloaded-neighbour hint without the sphere overload's
re-test, the destination box_intersects_cell gate with its deliberate
no-break, and add_all_outside_cells after the loop. The box-vs-cell BSP
traversal lands in BOTH representations behind the flat-authoritative
dispatcher with a graph referee whose 20,000 installed comparisons are
pinned by assertion (review F5), zero mismatch.
Dual Opus review: PASS on both lenses. The mandatory D0 pseudocode pass
caught that the contract's own supplementary note misattributed the box
block to the sphere overload — it belongs to a SECOND
check_building_transit overload @0x0052c680, whose portal-side
convention is INVERTED and whose admit differs; the pseudocode doc now
records that trap plus two byte confirmations made at review:
which_side @0x00444720 is strictly > eps for POSITIVE, and
intersect_box's in-plane early exit returns CROSSING(3)
(jp @0x005aa1bc -> mov eax,3), settling review items b1/b2 for the
future bridge porter. The bridge itself stays unported as AP-159's
explicit remainder.
The review also retired #335's severity premise honestly: "over-
inclusive only, never a missed one" is wrong at production shape ratios,
where the box (whole-vertex AABB) legitimately exceeds the sphere
(physics-polygon root sphere). Measured, both populations: rigged
(box << sphere) — 1,520 placements, 978 cells removed, 0 added;
production-ratio (box >= sphere) — 950 placements, 20 removed, 1 ADDED
through the loaded-neighbour gate, which is retail's direction, not a
defect. The no-op guard (review F4) asserts removal is nonzero so an
unwired admit cannot pass silently.
Process note: the implementer authored against this session's worktree
at bec5c69d, 25 commits stale — the recorded worktree-base class. All
six files were byte-identical between bases, the diff transplanted
losslessly, and every verdict-bearing run (referee, direction sweeps,
this clean-room) was re-executed on current main. S2's uncommitted
phase-1 edits were stashed for this landing so the suite verdicts
exactly one changeset.
Also untracks 341-slope-capture.jsonl (an accidental add) and
gitignores it.
Clean-room suite: 11,248 passed / 6 skipped / 0 failed.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Per docs/research/2026-07-30-ts4-116-oracle-plan.md §1, §4 item 2 (the
decisive TS-4 confirming run). Retail's BSP layer has NO steepness test at
all (acclient_2013_pseudo_c.txt:323783-323821, 0x0053a793) — every airborne
hit, steep or shallow, falls through to the same unconditional
`SetCollide` + `Adjusted`. The L.4 slide-tangent shortcut (worldNormal.Z <
FloorZ -> project-and-Slid, with its own SetSlidingNormal write) is deleted
from both BSPQuery.cs's and FlatBspQuery.cs's Path 6 sphere0 branch.
Fixing FlatBspQuery.cs (the flat/indexed engine Slice I6/I7 made
production-authoritative) was necessary in this same commit: it carried an
exact structural duplicate of the shortcut, caught by
FlatBspQueryDifferentialTests.InstalledDat_LargeRandomizedSweep_HasZeroBitMismatch
(graph=Adjusted vs flat=Slid) once the graph side was fixed alone. Its
sphere1 branch is also brought in line with the #116 shape-1 fix landed
in db2889af (direct Collided + SetCollisionNormal instead of the deferred
SetCollide/shortcut treatment) — that parity gap existed since shape-1's
commit only touched BSPQuery.cs and the randomized differential sweep
didn't happen to exercise the narrow foot-clear/head-hit case until this
session's broader change surfaced it.
DECISIVE CONFIRMING RUN (Ts4SteepRoofWedgeCaptureTests, per the plan's own
required test-first order): added
FallOntoSteepSlope_WithHorizontalVelocity_NeverFreezesForOverHalfASecond_AndReachesFloor
— the same steep-roof drop as the existing pure-vertical fixture, but with
a small residual horizontal velocity (vx=-0.3 m/s), matching the realistic
live-play input (WASD, jump momentum) that validated the shortcut on
2026-04-30. With the shortcut removed, this variant converges cleanly to
the flat floor with zero freeze. The pure-vertical fixture, run
shortcut-removed, DOES still freeze — per the oracle plan's root-cause
trace (§1.2 Step E), this is the DEGENERATE case: AdjustOffset's crease
projection (Cross(ContactPlane.Normal, SlidingNormal)) is mathematically
orthogonal to a purely-Z gravity offset, crushing it to zero every tick
before TransitionalInsert can run again — present identically in the raw
decomp, ACE's port, and this port. Renamed and re-asserted as a PINNED
known-degenerate test
(FallOntoSteepSlope_PureVertical_FreezesAtDegenerateFixedPoint_RetailParity)
rather than treated as a bug. Filed as register row AD-56.
BSPStepUpTests.C3_Path6_AirborneMoverHitsSteepSlope_ReturnsSlid pinned the
OLD shortcut's Slid-no-Collide behavior directly; renamed to
...ReturnsAdjustedAndSetsCollide and corrected to the retail-faithful
Adjusted/Collide=true outcome.
Register: TS-4 row retired (struck through, retirement note); AD-56 filed
for the pure-vertical degenerate case; the retire-next shortlist's TS-4
entry removed and renumbered.
Full AcDream.Core.Tests suite: 4060 passed / 2 skipped (D4 stays Skip-tagged
in this commit; its own un-skip is a separate, dependent test-only commit
for #116 shape-2), no regressions.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Port every current containment, overlap, walkable, and six-path moving-collision query to immutable integer-indexed assets behind a graph-authoritative referee. Exact synthetic, installed-DAT, complete-resolver, and zero-allocation gates prove bit-identical behavior before connected dual publication.
Co-authored-by: OpenAI Codex <codex@openai.com>
