jpmschweitzer/settled-reach
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

gitignore fix and ticket bookkeeping

Browse Source
This commit is contained in:
jpmschweitzer
2026-07-02 11:28:49 +02:00
parent 677de42bd2
commit 9753fcb48e
3 changed files with 28 additions and 0 deletions
Download Patch File Download Diff File Expand all files Collapse all files
.gitignore
+2
View File
@@ -114,3 +114,5 @@ wiki/economics/corporations/generated_corporations.toml
client/settings.db
client/settings.db-shm
client/settings.db-wal
.claude/settings.local.json
.pql/changelog/ticket_history/2026-07.sql
+13
View File
@@ -0,0 +1,13 @@
INSERT INTO ticket_history (ticket_record_id, field, old_value, new_value, changed_by, changed_at, created_at, updated_at, deleted_at, hash, canonical_version) VALUES ('06FGSK19FGZAC3H25VRAV4JRFW', 'description', 'A third believability lever, distinct from T-1080 (macro field gradient) and T-1081 (relief): within a SINGLE terrain class the sub-chunk seed should paint a spatially-coherent mosaic of micro-habitats, generalized to ALL terrain types. Motivating case: the probe''s ''Wetland 100%'' should resolve into clearings, copses, creeks, meadows, bogs, marshes — and likewise for every other class. NOT D-210/T-948: D-210 SubBiomeVariant is a MACRO biome tag per attractor (Tundra vs TemperateForest vs Wetland) — ''which biome''. THIS is finer: the intra-class MICRO-mosaic (''texture within a class''), at sub-chunk (<64 m) scale. MECHANIC: district/biome picks the PALETTE; the sub-chunk seed (SeedDomain::Voxel) paints the MOSAIC, reusing the derive-don''t-store tier (detail_scatter.rs, domain_warp, D-227). Three properties: (1) spatially COHERENT — patches (copse/clearing/bog/meadow) are blobs via low-frequency noise; creeks are lines following micro-drainage; never per-tile salt-and-pepper. (2) causally CONDITIONED on micro-relief/micro-drainage/moisture/slope/aspect (bogs in wettest lows, meadows on drier rises, copses where drainage improves, creeks along micro-thalwegs) — legible, not random. (3) MODULATES the existing D-228 axes, not a new axis. PER-CLASS VOCAB (one mechanism, per-class data): Wetland: clearing/copse/creek/meadow/bog/marsh; Forest: glade/dense stand/deadfall/brook/fern undergrowth; Grassland: tussock/scrub island/wildflower/dry wash; Sand: dune crest/slack/hardpan/oasis/scrub clump; Rock: outcrop/scree/ledge/lichen/talus; Gravel: bar/active channel/vegetated island; Lava: fresh sheet/weathered crust/tube collapse/pioneer scrub. Needs a Q/D-record before implementation. Relates to T-948 (D-210 macro), T-1081 (relief, a conditioning input), T-1083 (protocol must measure intra-class mosaic diversity).', 'A third believability lever, distinct from T-1080 (macro field gradient) and T-1081 (relief): within a SINGLE terrain class the sub-chunk seed should paint a spatially-coherent mosaic of micro-habitats, generalized to ALL terrain types. Motivating case: the probe''s ''Wetland 100%'' should resolve into clearings, copses, creeks, meadows, bogs, marshes — and likewise for every other class. NOT D-210/T-948: D-210 SubBiomeVariant is a MACRO biome tag per attractor (Tundra vs TemperateForest vs Wetland) — ''which biome''. THIS is finer: the intra-class MICRO-mosaic (''texture within a class''), at sub-chunk (<64 m) scale. MECHANIC: district/biome picks the PALETTE; the sub-chunk seed (SeedDomain::Voxel) paints the MOSAIC, reusing the derive-don''t-store tier (detail_scatter.rs, domain_warp, D-227). Three properties: (1) spatially COHERENT — patches (copse/clearing/bog/meadow) are blobs via low-frequency noise; creeks are lines following micro-drainage; never per-tile salt-and-pepper. (2) causally CONDITIONED on micro-relief/micro-drainage/moisture/slope/aspect (bogs in wettest lows, meadows on drier rises, copses where drainage improves, creeks along micro-thalwegs) — legible, not random. (3) MODULATES the existing D-228 axes, not a new axis. PER-CLASS VOCAB (one mechanism, per-class data): Wetland: clearing/copse/creek/meadow/bog/marsh; Forest: glade/dense stand/deadfall/brook/fern undergrowth; Grassland: tussock/scrub island/wildflower/dry wash; Sand: dune crest/slack/hardpan/oasis/scrub clump; Rock: outcrop/scree/ledge/lichen/talus; Gravel: bar/active channel/vegetated island; Lava: fresh sheet/weathered crust/tube collapse/pioneer scrub. Needs a Q/D-record before implementation. Relates to T-948 (D-210 macro), T-1081 (relief, a conditioning input), T-1083 (protocol must measure intra-class mosaic diversity).
---
Refinement (2026-07-02, /whats-next Si pass + user resolution): design gate resolved. First work item: write the Q/D-record (claim via `pql decisions claim D architecture`), pinning the five decisions below; then implement.
DECISIONS (user-resolved):
1. MECHANISM: new `voxel_mosaic()` octave band in detail_scatter.rs (~[64,32,16,8] m) + `SeedDomain::VoxelMosaic = 12` (append-only; update the pinned-discriminant test in server/src/seed.rs) → organic blob patches. Creeks + wet/dry conditioning follow the existing voxel_relief field''s micro-thalwegs (creeks along lowest-relief paths; bogs in lows, meadows on rises). ~4 extra value_noise calls/voxel — same cost class as voxel_relief.
2. VOCAB→AXES: expand D-228 enum VARIANTS (values on existing axes — not new axes) to cover the FULL shipped vocabulary NOW, all 7 classes. User rationale: palette composition affects the deterministic realization, so adding variants later would shift generated worlds — front-load the vocabulary. The complete axis-combination table for all 7×5–6 vocab items is the load-bearing content of the D-record, with a D-228 amendment for every item needing a new variant (deadfall / hardpan / lichen / fern-undergrowth have no honest existing mapping).
3. PALETTE RESIDENCY: `server/data/mosaic_constants.toml`, runtime-loaded like climate_constants.toml (T-1080 precedent). Note: the TOML is part of the deterministic input surface — weight changes change realizations; expected during Q-123 calibration, freeze after.
4. DoD METRIC IN SCOPE: add a within-patch intra-class variety field to ContrastMetrics (server/src/atlas/believability.rs) + a provisional criterion K≥3 distinct micro-features per sampled patch; Q-123 calibrates the final K. Matches the T-1081 precedent (mechanic + metric land together).
5. FAMILY GATING (grounded default, accepted): flat families (AlluvialPlain, LavaField, BraidedDelta, DuneStrand, MeanderReach) take the full mosaic; dramatic families (CliffCoast, FjordWall, IncisedGorge) get a per-family-gated non-structural modulation pass (vegetation/material only — must not warp carved geometry), mirroring voxel_relief gating. WaterBody: none. <5 ms chunk budget verification is part of implementation (T-1081 harness precedent).
CODE TOUCHPOINTS (Si-verified 2026-07-02): detail_scatter.rs:87-96 (voxel_relief pattern to extend), seed.rs:84-118 (SeedDomain; pinned-discriminant test ~:267), voxel.rs:1907-1943 (scatter_vegetation — the per-voxel salt-and-pepper this replaces), voxel.rs:445-495 (post-dispatch modulation seam), believability.rs (ContrastMetrics; evaluate_criteria ~:319). Overlaps to reconcile in the axis table: tube_depression (voxel.rs:710-717, Lava vocab) and braided-delta gravel-bar scrub (voxel.rs:1069-1092, Gravel vocab).', NULL, '2026-07-02 07:42:06', '2026-07-02 07:42:06.753', '2026-07-02 07:42:06.753', NULL, 'c8ddbe0307a798f866d712faa648eed6', 2) ON CONFLICT(hash) DO NOTHING;
.pql/changelog/tickets/2026-07.sql
+13
View File
@@ -0,0 +1,13 @@
INSERT INTO tickets (record_id, type, parent_record_id, title, description, status, priority, assigned_to, team, decision_ref, created_at, updated_at, deleted_at, hash, canonical_version) VALUES ('06FGSK19FGZAC3H25VRAV4JRFW', 'story', '06FGSJNP8YM537DJCC82ZD6544', 'Intra-class micro-habitat mosaic — seed-scattered sub-features within every terrain type', 'A third believability lever, distinct from T-1080 (macro field gradient) and T-1081 (relief): within a SINGLE terrain class the sub-chunk seed should paint a spatially-coherent mosaic of micro-habitats, generalized to ALL terrain types. Motivating case: the probe''s ''Wetland 100%'' should resolve into clearings, copses, creeks, meadows, bogs, marshes — and likewise for every other class. NOT D-210/T-948: D-210 SubBiomeVariant is a MACRO biome tag per attractor (Tundra vs TemperateForest vs Wetland) — ''which biome''. THIS is finer: the intra-class MICRO-mosaic (''texture within a class''), at sub-chunk (<64 m) scale. MECHANIC: district/biome picks the PALETTE; the sub-chunk seed (SeedDomain::Voxel) paints the MOSAIC, reusing the derive-don''t-store tier (detail_scatter.rs, domain_warp, D-227). Three properties: (1) spatially COHERENT — patches (copse/clearing/bog/meadow) are blobs via low-frequency noise; creeks are lines following micro-drainage; never per-tile salt-and-pepper. (2) causally CONDITIONED on micro-relief/micro-drainage/moisture/slope/aspect (bogs in wettest lows, meadows on drier rises, copses where drainage improves, creeks along micro-thalwegs) — legible, not random. (3) MODULATES the existing D-228 axes, not a new axis. PER-CLASS VOCAB (one mechanism, per-class data): Wetland: clearing/copse/creek/meadow/bog/marsh; Forest: glade/dense stand/deadfall/brook/fern undergrowth; Grassland: tussock/scrub island/wildflower/dry wash; Sand: dune crest/slack/hardpan/oasis/scrub clump; Rock: outcrop/scree/ledge/lichen/talus; Gravel: bar/active channel/vegetated island; Lava: fresh sheet/weathered crust/tube collapse/pioneer scrub. Needs a Q/D-record before implementation. Relates to T-948 (D-210 macro), T-1081 (relief, a conditioning input), T-1083 (protocol must measure intra-class mosaic diversity).
---
Refinement (2026-07-02, /whats-next Si pass + user resolution): design gate resolved. First work item: write the Q/D-record (claim via `pql decisions claim D architecture`), pinning the five decisions below; then implement.
DECISIONS (user-resolved):
1. MECHANISM: new `voxel_mosaic()` octave band in detail_scatter.rs (~[64,32,16,8] m) + `SeedDomain::VoxelMosaic = 12` (append-only; update the pinned-discriminant test in server/src/seed.rs) → organic blob patches. Creeks + wet/dry conditioning follow the existing voxel_relief field''s micro-thalwegs (creeks along lowest-relief paths; bogs in lows, meadows on rises). ~4 extra value_noise calls/voxel — same cost class as voxel_relief.
2. VOCAB→AXES: expand D-228 enum VARIANTS (values on existing axes — not new axes) to cover the FULL shipped vocabulary NOW, all 7 classes. User rationale: palette composition affects the deterministic realization, so adding variants later would shift generated worlds — front-load the vocabulary. The complete axis-combination table for all 7×5–6 vocab items is the load-bearing content of the D-record, with a D-228 amendment for every item needing a new variant (deadfall / hardpan / lichen / fern-undergrowth have no honest existing mapping).
3. PALETTE RESIDENCY: `server/data/mosaic_constants.toml`, runtime-loaded like climate_constants.toml (T-1080 precedent). Note: the TOML is part of the deterministic input surface — weight changes change realizations; expected during Q-123 calibration, freeze after.
4. DoD METRIC IN SCOPE: add a within-patch intra-class variety field to ContrastMetrics (server/src/atlas/believability.rs) + a provisional criterion K≥3 distinct micro-features per sampled patch; Q-123 calibrates the final K. Matches the T-1081 precedent (mechanic + metric land together).
5. FAMILY GATING (grounded default, accepted): flat families (AlluvialPlain, LavaField, BraidedDelta, DuneStrand, MeanderReach) take the full mosaic; dramatic families (CliffCoast, FjordWall, IncisedGorge) get a per-family-gated non-structural modulation pass (vegetation/material only — must not warp carved geometry), mirroring voxel_relief gating. WaterBody: none. <5 ms chunk budget verification is part of implementation (T-1081 harness precedent).
CODE TOUCHPOINTS (Si-verified 2026-07-02): detail_scatter.rs:87-96 (voxel_relief pattern to extend), seed.rs:84-118 (SeedDomain; pinned-discriminant test ~:267), voxel.rs:1907-1943 (scatter_vegetation — the per-voxel salt-and-pepper this replaces), voxel.rs:445-495 (post-dispatch modulation seam), believability.rs (ContrastMetrics; evaluate_criteria ~:319). Overlaps to reconcile in the axis table: tube_depression (voxel.rs:710-717, Lava vocab) and braided-delta gravel-bar scrub (voxel.rs:1069-1092, Gravel vocab).', 'backlog', 'high', NULL, 'server', 'D-227', '2026-06-28 05:55:34.396', '2026-07-02 07:42:06.753', NULL, '0cd112f1fc4f62f5b7a464dd51ae5205', 2) ON CONFLICT(record_id) DO UPDATE SET type=excluded.type, parent_record_id=excluded.parent_record_id, title=excluded.title, description=excluded.description, status=excluded.status, priority=excluded.priority, assigned_to=excluded.assigned_to, team=excluded.team, decision_ref=excluded.decision_ref, updated_at=excluded.updated_at, deleted_at=excluded.deleted_at, hash=excluded.hash, canonical_version=excluded.canonical_version WHERE excluded.updated_at > tickets.updated_at OR (excluded.updated_at = tickets.updated_at AND excluded.hash > tickets.hash);