code wiki / (root) / nx_world_quality_grader_gate.nx

nx_world_quality_grader_gate.nx

buildroot/runtime/nx_world_quality_grader_gate.nx

15753 B263 linesdepth 3pulls 4 transitivereach 0 importersview sourcekind gate/prooftopic world
docsdependenciesstructsconstsfunctions

about

nx_world_quality_grader_gate.nx -- THE REFEREE FOR THE WORLD REFEREE, 2026-09-01. nx_world_quality_grader.nx shipped on 2026-05-16 and was NEVER BUILT: catalog read SOURCE-ONLY (S2) for it and for every one of its five consumers (nx_procgen_biome_test, nx_procgen_water_test, nx_procgen_features_test, nx_procgen_signature_test, nx_layer_verdict). A grader that has never compiled has never graded anything, and the estate law for that is a build-and-run deficit, not a cleverness deficit. IN-PROCESS over nx_world_quality_grader.nx, deliberately. A gate that fork/exec'd a deployed elf would report NOT-REACHED for every mutant and its GREEN would prove nothing about the code under test. WHAT THIS PROVES, STATED SO IT CANNOT BE OVER-READ: the 8-axis decision core -- the degenerate-input contract, the flatness INVERSION, the banded local-variation verdict, the optional-axis abstention, the biome and feature-variety rulers, the letter rubric and the refine-axis enum. It does NOT prove that the axes correlate with a human judgement of a world beauty; that is a referee-panel question and is not claimed here. THE SELF-TEST IN THE GRADER OWN main() IS NOT THIS GATE. It has no negative controls, no denominators, and its exit code is a bare 0 -- nothing downstream can read a verdict from it. Every tooth below that the self-test also covers is restated here so the verdict travels in the EXIT CODE, and the ones it never had are the reason this file exists. FIXTURES ARE BUILT IN MEMORY ON EVERY RUN AND TOUCH NO PATH AT ALL, so the run is idempotent by construction and can never share a fixture with a production beat. license_tier: ORIGINAL No hw writes (Rule 26). expect_exit: 0

dependencies 2 imports · 0 importers

nx_gate_verdict.nx nx_world_quality_grader.nx nx_world_quality_grader_gate.n

imports: nx_gate_verdict.nxnx_world_quality_grader.nx

imported by: nobody (leaf or entry point)

structs

none

consts

30const WG_W: i64 = 8 // the self-test own grid, so outcomes stay comparable
31const WG_H: i64 = 8
32const WG_RELIEF: i64 = 1000 // max_expected_relief handed to every grade below
33const WG_FLAT_VALUE: i64 = 100 // the constant a degenerate world is filled with
34const WG_MAPS_GRADED: i64 = 7 // DENOMINATOR: grades this run must actually perform
35const WG_PEAK: i64 = 1000 // mountain apex before the radial falloff
36const WG_FALLOFF: i64 = 50 // radial falloff per unit squared distance
37const WG_JITTER: i64 = 50 // checkerboard jitter amplitude on the mountain map
38const WG_CENTRE: i64 = 4 // WG_W/2: the mountain apex column
39const WG_BIOME_DESERT: i64 = 8 // expected band LOW
40const WG_BIOME_TFOREST: i64 = 5 // expected band MID
41const WG_BIOME_SNOW: i64 = 12 // expected band HIGH
42const WG_BIOME_ICE: i64 = 13 // expected band PEAK
43const WG_KINDS: i64 = 4 // feature-kind histogram width
44const WG_KIND_COUNT: i64 = 10 // per-kind count in the UNIFORM histogram
45const WG_KIND_ALL: i64 = 40 // WG_KINDS * WG_KIND_COUNT, all in one kind
46const WG_QUARTER: i64 = 4 // the grader own elevation-band divisor
47const WG_TWO: i64 = 2
48const WG_THREE: i64 = 3

functions

50func wg_scan_min(m: *i64, n: i64) -> i64
60func wg_scan_max(m: *i64, n: i64) -> i64
70func main(argc: i64, argv: *i64) -> i64