The batch's real bug: extract_river_network computed the seaward
neighbor (nr,nc) to decide the MOUTH sentinel then discarded it, and
build_edges placeholder-pointed mouth edges at themselves — zero
chord, invent_course's degenerate 1-point return, resolve_mouth_
terminus dead code on real data, ALL real mouths resolving None, and
(because Ruling 3g retired the D/Q clip on the promise of real
termini) mouths vanishing at District/Quarter. The second instance of
the threw-away-the-answer anti-pattern Ruling 2b fixed for interior
pointers. Fix: river_seaward: Vec<(u16,u16)> on RiverNetwork
(additive, serde-default, parallel array; meaningful only at MOUTH
entries), captured in the same extraction pass; build_edges gives
Mouth edges the real one-D8-step chord. Permanent acceptance:
all_real_gj1c_mouths_resolve_to_mouth_terminus_not_none — 3/3, revert-
verified failing at the golden's first mouth (38,47). Mouth golden
coverage added (river_course_golden gains district_mouth/quarter_mouth
samples + non-degeneracy test; the previously Interior-only filter
gap closed). Tyre 1: the land-probe's dead dx/len*len arithmetic
replaced — station_spacing_m threaded through the crop path, probe
steps one real cell spacing, comment reconciled. Tyre 2: boxing
comment reattributed to variant-size balancing (Layer1Output retention
lives on TerrainAnalysisCache, not BodyWorldState). Hoshe #4:
cascade_golden's doc now states the attractor cascade accurately
(count-parity, not byte-identity — water_dist seeds from mouths).
Full cargo test green; goldens re-pinned deliberately; bench +3.0%.
Tickets: T-1170
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
A2 (river_course.rs): Stage A rung-independent valley-seeking control
path (chord/8 stations, k=5 bilinear-scored candidates + continuity
penalty); Stage B rung-indexed perpendicular warp on GLOBAL arc-length
(window-independence, Ruling 1e), band chord/2 down to min_wavelength_m
hard-truncate, sine taper to zero at anchors, amplitude min(8% chord,
half-cell) slope/class-scaled. SeedDomain::RiverCourse=17, distinct
salt. Wire: RiverCourse{edge_id,class,points,terminus} on
DistrictWindowLayer.courses (serde-default); bbox-culled, window-
cropped +1 station. TerrainAnalysisCache retains Layer1Output (the
gen_queue:626 discard, Ruling 4b). A3: mouth termination walks
stations sampling the window's OWN rung-consistent morphology verdict,
6-iteration bisect; land-at-anchor probes one segment then None;
EdgeDrain never probes. A5: near_perennial_water point-to-segment
predicate (D-239 §8 governed bands 1-3m class-scaled) threaded through
both batch and window paths; Region always false; never touches
moisture_q. Discipline closed: dormant zoom-ladder bench run + numbers
recorded in the design doc (District 3.009/1.785, Quarter 1.823
us/cell, Region window 0.617ms); course-cost bench CAUGHT a real
+12-36% per-cell riparian scan regression -> precomputed bbox O(1)
reject (60ns->2.2ns/call), final delta +3.4-6.9% at budget; three
determinism tests (overlapping-window byte-identity, cross-rung
amplitude bound, warp-stream cross-correlation r<0.3); goldens: window
sweep gained a verified course-bearing position (pure append), new
river_course golden at both rungs, believability verified unchanged.
Revert-verification discovered the pole-row branch is structurally
unreachable (flow_direction bounds-check) — the real edge-drain path
is k<0 flat-plateau; test fixture rewritten to exercise reality.
scale.rs stale comment fixed. Full cargo test green.
Tickets: T-1170, T-1168
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
jpmschweitzerandClaude Fable 5