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

feat(db): corp-HQ settlement model + population/class bake (D-242) #177

Closed
jpmschweitzer wants to merge 0 commits from settlement-dataset-d242 into main
pull from: settlement-dataset-d242
merge into: :main
Conversation 3 Commits - Files Changed -
jpmschweitzer commented 2026-07-16 11:49:18 +02:00
Owner
Copy Link
Copy Source

Summary

Implements D-242 end-to-end: corp HQs leave the city pool, and per-settlement population + settlement_class are baked at import. Closes T-1074 + T-1075; unblocks T-1076 (road-graph hub refinement).

T-1074 — corp-HQ settlement model

  • Removed the reserved=1 corp-HQ cross-reference (populate_atlas_city_names_corps) — all 30 duplicate co-named city groups (10x Groombridge on GJ380c, etc.) eliminated; 0 duplicate (body_id, name) groups post-regen.
  • New corporations.corp_specialization keyed on the D-237 vocabulary, extended 27 -> 30 (trade_distribution, hospitality_hub, professional_services); authored per corp as wiki frontmatter across 155 pages.
  • Authored wiki/economics/corp_hq_placement.toml: all 30 values -> {CityTenant, Standalone} + standalone_economic_role.
  • CityTenant: new headquarters_city_id FK (58/60 linked). Standalone: 94 settlements emitted as ordinary atlas_city_names rows — zero Rust production changes at the read_body_settlements -> match_cities seam; 2 new regression tests lock the invariant.
  • headquarters_body backfilled via the lifted most-populated-body heuristic + body-type tiebreak fix (154/155; the 1 gap is a pre-existing broken Sova ref — T-1054).
  • Retired the orphaned tooling/populate-corporations.sh hand-seed script.

T-1075 — population/class bake

  • import_economics derives per-city population from bodies.population via a documented integer Zipf rank-size curve on every regen — sums exactly, 0 discrepancies across 276 bodies.
  • settlement_class defaults PopulationBudget (D-196 thresholds); wiki/economics/settlement_name_locked.toml pins 8 hero cities NameLocked. EconomicTriggered/OrganicGrowth stay simulation-time.

Governance

D-242 implementation status + derivation-at-import amendment; D-207/D-223 supersede notes; new TOMLs registered in tooling/generator_sources.py.

Verification

3x consecutive make regen-db deterministic (identical counts), stamp fresh; full cargo test green; clippy --all-targets -D warnings clean; ruff + make test-tooling green; live cascade run places a Standalone HQ alongside pooled cities through the unmodified pipeline.

Follow-ups

T-1115 filed: 10 wiki-pageless legacy corps + GJ 702B has zero authored city names.

## Summary Implements D-242 end-to-end: corp HQs leave the city pool, and per-settlement population + settlement_class are baked at import. Closes T-1074 + T-1075; unblocks T-1076 (road-graph hub refinement). ### T-1074 — corp-HQ settlement model - Removed the reserved=1 corp-HQ cross-reference (`populate_atlas_city_names_corps`) — all 30 duplicate co-named city groups (10x Groombridge on GJ380c, etc.) eliminated; 0 duplicate (body_id, name) groups post-regen. - New `corporations.corp_specialization` keyed on the D-237 vocabulary, extended 27 -> 30 (trade_distribution, hospitality_hub, professional_services); authored per corp as wiki frontmatter across 155 pages. - Authored `wiki/economics/corp_hq_placement.toml`: all 30 values -> {CityTenant, Standalone} + standalone_economic_role. - CityTenant: new `headquarters_city_id` FK (58/60 linked). Standalone: 94 settlements emitted as ordinary `atlas_city_names` rows — zero Rust production changes at the `read_body_settlements` -> `match_cities` seam; 2 new regression tests lock the invariant. - `headquarters_body` backfilled via the lifted most-populated-body heuristic + body-type tiebreak fix (154/155; the 1 gap is a pre-existing broken Sova ref — T-1054). - Retired the orphaned `tooling/populate-corporations.sh` hand-seed script. ### T-1075 — population/class bake - `import_economics` derives per-city population from `bodies.population` via a documented integer Zipf rank-size curve on every regen — sums exactly, 0 discrepancies across 276 bodies. - `settlement_class` defaults PopulationBudget (D-196 thresholds); `wiki/economics/settlement_name_locked.toml` pins 8 hero cities NameLocked. EconomicTriggered/OrganicGrowth stay simulation-time. ### Governance D-242 implementation status + derivation-at-import amendment; D-207/D-223 supersede notes; new TOMLs registered in `tooling/generator_sources.py`. ### Verification 3x consecutive `make regen-db` deterministic (identical counts), stamp fresh; full cargo test green; clippy --all-targets -D warnings clean; ruff + make test-tooling green; live cascade run places a Standalone HQ alongside pooled cities through the unmodified pipeline. ### Follow-ups T-1115 filed: 10 wiki-pageless legacy corps + GJ 702B has zero authored city names.
jpmschweitzer added 6 commits 2026-07-16 11:49:18 +02:00
chore(meta): activate T-1074/T-1075 settlement-dataset batch — Si refinement + gap resolutions appended 1f958962cc
feat(db): backfill corp_specialization frontmatter across 155 corp wiki pages (T-1074, D-242) 43a6d55180 
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
feat(db): corp-HQ settlement model + per-settlement population/class bake (T-1074/T-1075, D-242) 62bbf57009 
Remove the reserved=1 corp-HQ city-pool cross-reference (duplicate co-named cities); corp_specialization keyed on the D-237 vocabulary extended 27->30; authored corp_hq_placement.toml {CityTenant,Standalone} map; headquarters_body backfill via lifted most-populated-body heuristic + body-type tiebreak; standalone HQs emitted as ordinary settlement rows; Zipf rank-size population spread from bodies.population at import; settlement_class defaults PopulationBudget + settlement_name_locked.toml hero pins; retire orphaned populate-corporations.sh; regen systems.db.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
test(simulation): lock standalone-HQ-as-ordinary-settlement-row invariant (T-1074) 314894d080 
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
docs(meta): D-242 implementation status + derivation amendment; D-207/D-223 supersede notes (T-1074/T-1075) c0de8ce68c 
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
style(simulation): cargo fmt — test fixture tuple layout 0c8a270471 
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
jpmschweitzer commented 2026-07-16 12:24:48 +02:00
Author
Owner
Copy Link
Copy Source

Review: settlement-dataset-d242 → main (type: code + content slice)

Reviewers: Hoshe (code quality), Tyre (architecture), Miri (world consistency — wiki/vocabulary slice). All three read source from the branch worktree (HEAD 0c8a27047, verified = branch tip, not sparse). Pre-push gate (fmt/clippy/full cargo test/deny/ruff/test-tooling/stamp) passed before push and was not duplicated here. Runtime verification: no client or merge-path code touched; live cascade run documented in the PR description.

Hoshe (Code Quality): REQUEST_CHANGES

Zipf population-spread math verified correct (exact-sum, non-negative, monotonic across 20,000+ randomized trials), Rust test additions genuinely load-bearing, SQL parameterization clean throughout. But the headquarters_body backfill has a live bug, and the tiebreak it feeds rests on a premise the PR's own data disproves.

# File Issue
H1 tooling/economy-db/economy_import/corporations.py:292-313 headquarters_body backfill is a permanent one-shot no-op: corporations is append-only across regen runs and the if hq_body: continue guard treats the heuristic's own prior output as authored data (nothing is actually authored in this column). Live-verified: 0 headquarters_body backfilled on every rerun; prometheus-labs/kovalev-freight are stuck on GJ702B-belt while the branch's fixed heuristic returns GJ702Bb. Lead independently confirmed against the committed DB. Fix: provenance marker distinguishing authored vs derived (so derived always re-runs), or drop the guard until an authoring path exists.
H2 tooling/economy-db/economy_import/atlas.py:167-184 _BODY_TYPE_SETTLEMENT_PREFERENCE premise ("asteroid_belt/oort_cloud are geometry, never named settlements") is contradicted by this PR's own generated data in 8 of 9 divergence cases — populate_standalone_hq_settlements freely places settlements on belts (GJ845-belt → Orkney Ceramics, GJ268-belt → Jeju Lattice, GJ222A-belt → Groot Karoo Cellars + Kalahari Leatherworks). If H1 is fixed as-is, the tiebreak would forcibly move calluna-wellness and namsan-collective off currently-working belt tenant links onto planets with zero cities. Either the type-ranking should defer to "does this body have city rows", or the standalone-HQ insertion needs the same body-type restriction. H1+H2 need one joint design call.
H3 governance/decisions/architecture.md (D-242 implementation-status entry) + wiki/economics/settlement_name_locked.toml header The "verified against a real, repeated make regen-db / deterministic across 3 consecutive runs" claims describe a no-op being stably a no-op — the backfill+tiebreak never actually fire after the first run. Correct the verification note once H1/H2 land so the next reader doesn't assume the tiebreak is exercised in production.

Tyre (Architecture): REQUEST_CHANGES

Implementation cross-checked against D-242, D-237, D-223, D-207, D-211, D-196, D-010 and the asset-pipeline rule. Determinism of the pure-integer Zipf spread holds; D-242 verification counts match the committed systems.db exactly; vocabulary extension is clean 1:1 coverage; the FK-clear-ordering fix is correct; "standalone HQ = ordinary row" tradeoff is explicitly disclosed in D-242 and works on current data (checked: 94/95 recoverable via name join, zero collisions) — not flagged. Two gaps:

# File Issue
T1 tooling/generator_sources.py:134-152 wiki/corporations/*.md (155 files) is not in the stamp source set, but these pages now author corp_specialization → drives hq_placement, headquarters_body, and standalone settlement-row creation. An edit there without make regen-db slips past check-systems-db-stamp and the pre-push hook, contradicting the asset-pipeline freshness guarantee. Pre-existing gap, but this PR makes it load-bearing. Fix: glob CORPORATIONS_DIR into IMPORT_ECONOMICS_SOURCES.
T2 tooling/economy-db/economy_import/corporations.py:284-291 populate_corp_specialization only UPDATEs corps present in the current authored set — no NULL-reset pass first, so a removed frontmatter key leaves a stale baked value and rebuild-from-source determinism breaks on import history. The sibling settlement_class bake (atlas.py:333-337, this same PR) implements reset-before-override correctly for exactly this reason. Mirror it. (headquarters_body's never-clobber is a separate, intentional design — see H1 for its own problem.)

Miri (World Consistency): REQUEST_CHANGES

Sampled 36/155 corp pages (all five flagged judgment calls + every specialization bucket), full re-read of the vocabulary/placement/hero-pin TOMLs, cross-checked against D-237 examples and system wiki prose. Categorization accurate in every sampled case but one; the 3 new vocabulary values, all 30 placement mappings, and all 8 NameLocked pins check out.

# File Issue
M1 wiki/corporations/somatic-futures.md (corp_specialization: professional_services) Miscategorized. The page describes "Licensed Re-embodiment and Longevity Services", "the Assembly re-embodiment medical license", "40+ clinic locations" — a licensed clinical-services institution central to the setting's longevity/Perpetual-class lore, not general advisory (professional_services anchors on insurance/consulting/arbitration). longevity_monopoly is correctly reserved for Prometheus Labs (sole provider); Somatic Futures is explicitly non-monopoly ("and a small number of other certified operators"). This is a second instance of the exact vocabulary-gap class this PR fixed with its 3 new values. Fix: add licensed_clinical_services (anchor medical_services, NonPhysical → CityTenant, Specialist — non-monopoly counterpart to longevity_monopoly) and re-tag somatic-futures. Placement (CityTenant at Kallast) is correct either way.

Verdict: CHANGES REQUESTED

6 findings (H1-H3, T1-T2, M1). H1+H2 are one joint design decision; T1/T2/M1 are independent, well-scoped fixes. H3 is a documentation correction downstream of H1/H2.

Process note: a reviewer's live-import verification dirtied the worktree systems.db; the lead restored it to the committed state before this comment — the branch tree is clean for the fix round.

## Review: settlement-dataset-d242 → main (type: code + content slice) Reviewers: Hoshe (code quality), Tyre (architecture), Miri (world consistency — wiki/vocabulary slice). All three read source from the branch worktree (HEAD `0c8a27047`, verified = branch tip, not sparse). Pre-push gate (fmt/clippy/full cargo test/deny/ruff/test-tooling/stamp) passed before push and was not duplicated here. Runtime verification: no client or merge-path code touched; live cascade run documented in the PR description. ### Hoshe (Code Quality): REQUEST_CHANGES Zipf population-spread math verified correct (exact-sum, non-negative, monotonic across 20,000+ randomized trials), Rust test additions genuinely load-bearing, SQL parameterization clean throughout. But the headquarters_body backfill has a live bug, and the tiebreak it feeds rests on a premise the PR's own data disproves. | # | File | Issue | |---|------|-------| | H1 | tooling/economy-db/economy_import/corporations.py:292-313 | headquarters_body backfill is a **permanent one-shot no-op**: `corporations` is append-only across regen runs and the `if hq_body: continue` guard treats the heuristic's own prior output as authored data (nothing is actually authored in this column). Live-verified: `0 headquarters_body backfilled` on every rerun; prometheus-labs/kovalev-freight are stuck on `GJ702B-belt` while the branch's fixed heuristic returns `GJ702Bb`. Lead independently confirmed against the committed DB. Fix: provenance marker distinguishing authored vs derived (so derived always re-runs), or drop the guard until an authoring path exists. | | H2 | tooling/economy-db/economy_import/atlas.py:167-184 | `_BODY_TYPE_SETTLEMENT_PREFERENCE` premise ("asteroid_belt/oort_cloud are geometry, never named settlements") is contradicted by this PR's own generated data in 8 of 9 divergence cases — populate_standalone_hq_settlements freely places settlements on belts (GJ845-belt → Orkney Ceramics, GJ268-belt → Jeju Lattice, GJ222A-belt → Groot Karoo Cellars + Kalahari Leatherworks). If H1 is fixed as-is, the tiebreak would forcibly move calluna-wellness and namsan-collective off currently-working belt tenant links onto planets with zero cities. Either the type-ranking should defer to "does this body have city rows", or the standalone-HQ insertion needs the same body-type restriction. H1+H2 need one joint design call. | | H3 | governance/decisions/architecture.md (D-242 implementation-status entry) + wiki/economics/settlement_name_locked.toml header | The "verified against a real, repeated make regen-db / deterministic across 3 consecutive runs" claims describe a no-op being stably a no-op — the backfill+tiebreak never actually fire after the first run. Correct the verification note once H1/H2 land so the next reader doesn't assume the tiebreak is exercised in production. | ### Tyre (Architecture): REQUEST_CHANGES Implementation cross-checked against D-242, D-237, D-223, D-207, D-211, D-196, D-010 and the asset-pipeline rule. Determinism of the pure-integer Zipf spread holds; D-242 verification counts match the committed systems.db exactly; vocabulary extension is clean 1:1 coverage; the FK-clear-ordering fix is correct; "standalone HQ = ordinary row" tradeoff is explicitly disclosed in D-242 and works on current data (checked: 94/95 recoverable via name join, zero collisions) — not flagged. Two gaps: | # | File | Issue | |---|------|-------| | T1 | tooling/generator_sources.py:134-152 | `wiki/corporations/*.md` (155 files) is not in the stamp source set, but these pages now author corp_specialization → drives hq_placement, headquarters_body, and standalone settlement-row creation. An edit there without `make regen-db` slips past check-systems-db-stamp and the pre-push hook, contradicting the asset-pipeline freshness guarantee. Pre-existing gap, but this PR makes it load-bearing. Fix: glob CORPORATIONS_DIR into IMPORT_ECONOMICS_SOURCES. | | T2 | tooling/economy-db/economy_import/corporations.py:284-291 | populate_corp_specialization only UPDATEs corps present in the current authored set — no NULL-reset pass first, so a removed frontmatter key leaves a stale baked value and rebuild-from-source determinism breaks on import history. The sibling settlement_class bake (atlas.py:333-337, this same PR) implements reset-before-override correctly for exactly this reason. Mirror it. (headquarters_body's never-clobber is a separate, intentional design — see H1 for its own problem.) | ### Miri (World Consistency): REQUEST_CHANGES Sampled 36/155 corp pages (all five flagged judgment calls + every specialization bucket), full re-read of the vocabulary/placement/hero-pin TOMLs, cross-checked against D-237 examples and system wiki prose. Categorization accurate in every sampled case but one; the 3 new vocabulary values, all 30 placement mappings, and all 8 NameLocked pins check out. | # | File | Issue | |---|------|-------| | M1 | wiki/corporations/somatic-futures.md (corp_specialization: professional_services) | Miscategorized. The page describes "Licensed Re-embodiment and Longevity Services", "the Assembly re-embodiment medical license", "40+ clinic locations" — a licensed clinical-services institution central to the setting's longevity/Perpetual-class lore, not general advisory (professional_services anchors on insurance/consulting/arbitration). longevity_monopoly is correctly reserved for Prometheus Labs (sole provider); Somatic Futures is explicitly non-monopoly ("and a small number of other certified operators"). This is a second instance of the exact vocabulary-gap class this PR fixed with its 3 new values. Fix: add `licensed_clinical_services` (anchor medical_services, NonPhysical → CityTenant, Specialist — non-monopoly counterpart to longevity_monopoly) and re-tag somatic-futures. Placement (CityTenant at Kallast) is correct either way. | ### Verdict: CHANGES REQUESTED 6 findings (H1-H3, T1-T2, M1). H1+H2 are one joint design decision; T1/T2/M1 are independent, well-scoped fixes. H3 is a documentation correction downstream of H1/H2. Process note: a reviewer's live-import verification dirtied the worktree systems.db; the lead restored it to the committed state before this comment — the branch tree is clean for the fix round.
jpmschweitzer added 2 commits 2026-07-16 13:02:42 +02:00
fix(db): PR #177 review round — H1/H2/T1/T2/M1 e63caeb620 
H1+H2: headquarters_body is reset+derived every run (DB is never source); optional authored frontmatter override, hard-validated; tiebreak now population DESC -> city-bearing body -> type rank -> body_id, preserving belt tenancies and fixing GJ702B to GJ702Bb. T2: NULL-reset pass for corp_specialization/hq_placement before authored re-apply (poison-tested). T1: wiki/corporations/*.md globbed into IMPORT_ECONOMICS_SOURCES. M1: licensed_clinical_services vocabulary value (31st, NonPhysical->CityTenant) + somatic-futures retag. Fixpoint verified stable across 4 consecutive regens (0 diffs); regen systems.db.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
docs(meta): H3 — correct D-242 verification wording to recompute-every-run behavior 9bb533a8b0 
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
jpmschweitzer commented 2026-07-16 13:02:52 +02:00
Author
Owner
Copy Link
Copy Source

Review response — all 6 findings fixed (commits e63caeb62, 9bb533a8b)

No pushbacks; every finding was accepted and addressed. Lead independently re-verified the headline claims against the regenerated DB before this comment.

# Finding Resolution
H1 headquarters_body backfill permanent one-shot no-op Fixed — reset+derive every run. The three importer-owned columns (corp_specialization, hq_placement, headquarters_body) are NULLed for all corps each run, authored values re-applied from source, heuristic re-resolves the rest. An optional authored headquarters_body: frontmatter override is implemented now (hard-validated: body must exist and sit in the corp's HQ system) — authored data lives in source files, never in DB state. Verified: 4 consecutive regens each print "154 headquarters_body resolved (recomputed every run)" with full-snapshot diffs A=B=C=D at 0 differences; a poisoned-DB test (garbage in all three columns) fully re-derives a wiki corp and reverts a legacy corp to NULL.
H2 Body-type tiebreak premise contradicted by belt settlements Fixed — city-presence tier added. Tiebreak is now population DESC → city-bearing body (has atlas_city_names rows) → body-type rank → body_id. Both working belt tenancies are preserved with identical targets (calluna-wellness → Orkney Ceramics on GJ845-belt; namsan-collective → Jeju Lattice on GJ268-belt — verified the only signal that can preserve these is the presence tier, since those settlements are themselves Standalone-HQ rows). GJ702B corps move to planet GJ702Bb (only 2 placement changes repo-wide vs baseline; still unmatched tenants pending T-1115, logged not errored). The presence set reads the previous run's settled atlas_city_names state — documented as deliberate hysteresis (one-run lag for brand-new Standalone corps; loud warning if empty) in the docstring and D-242, with the fixpoint proven stable empirically. _BODY_TYPE_SETTLEMENT_PREFERENCE comment rewritten to the corrected premise.
H3 Verification wording describes a no-op being stably a no-op Fixed. D-242 status entry and settlement_name_locked.toml header rewritten to describe recompute-every-run behavior and the actually-measured outcomes (incl. why the 3 corp-named NameLocked pins survive recompute: population-first re-derivation, not held-over state). Prometheus gap now cites T-1115.
T1 Corp wiki pages not in the stamp source set Fixed. CORPORATIONS_DIR moved into the registry; _corporation_pages() globs wiki/corporations/*.md (sorted, index.md excluded, fail-closed) into IMPORT_ECONOMICS_SOURCES; paths.py re-exports from the registry. check-systems-db confirms the stamp now covers the 155 pages.
T2 No reset pass for corp_specialization/hq_placement Fixed. NULL-reset for all corps before the authored re-apply, mirroring the settlement_class reset-before-override pattern. Removed-key case verified: a corp dropped from the authored set reverts to all-NULL on the next run.
M1 somatic-futures miscategorized as professional_services Fixed. New licensed_clinical_services vocabulary value (31st: anchor medical_services, NonPhysical, Specialist, override concentrated — passes V-SES-03), corp_hq_placement.toml stanza (CityTenant; no standalone_economic_role per the file's Standalone-only convention), somatic-futures retagged. Placement unchanged (CityTenant at Kallast, GJ144d). 155/155 specialized; vocabulary 31 values / placement map 31 stanzas.

Gate re-run on push: fmt/clippy/full cargo test/deny/ruff/test-tooling/stamp all green. Reviewers have been asked to verify their own findings for the re-review verdict.

## Review response — all 6 findings fixed (commits e63caeb62, 9bb533a8b) No pushbacks; every finding was accepted and addressed. Lead independently re-verified the headline claims against the regenerated DB before this comment. | # | Finding | Resolution | |---|---------|------------| | H1 | headquarters_body backfill permanent one-shot no-op | **Fixed — reset+derive every run.** The three importer-owned columns (corp_specialization, hq_placement, headquarters_body) are NULLed for all corps each run, authored values re-applied from source, heuristic re-resolves the rest. An optional authored `headquarters_body:` frontmatter override is implemented now (hard-validated: body must exist and sit in the corp's HQ system) — authored data lives in source files, never in DB state. Verified: 4 consecutive regens each print "154 headquarters_body resolved (recomputed every run)" with full-snapshot diffs A=B=C=D at 0 differences; a poisoned-DB test (garbage in all three columns) fully re-derives a wiki corp and reverts a legacy corp to NULL. | | H2 | Body-type tiebreak premise contradicted by belt settlements | **Fixed — city-presence tier added.** Tiebreak is now population DESC → **city-bearing body** (has atlas_city_names rows) → body-type rank → body_id. Both working belt tenancies are preserved with identical targets (calluna-wellness → Orkney Ceramics on GJ845-belt; namsan-collective → Jeju Lattice on GJ268-belt — verified the only signal that can preserve these is the presence tier, since those settlements are themselves Standalone-HQ rows). GJ702B corps move to planet GJ702Bb (only 2 placement changes repo-wide vs baseline; still unmatched tenants pending T-1115, logged not errored). The presence set reads the previous run's settled atlas_city_names state — documented as deliberate hysteresis (one-run lag for brand-new Standalone corps; loud warning if empty) in the docstring and D-242, with the fixpoint proven stable empirically. _BODY_TYPE_SETTLEMENT_PREFERENCE comment rewritten to the corrected premise. | | H3 | Verification wording describes a no-op being stably a no-op | **Fixed.** D-242 status entry and settlement_name_locked.toml header rewritten to describe recompute-every-run behavior and the actually-measured outcomes (incl. why the 3 corp-named NameLocked pins survive recompute: population-first re-derivation, not held-over state). Prometheus gap now cites T-1115. | | T1 | Corp wiki pages not in the stamp source set | **Fixed.** CORPORATIONS_DIR moved into the registry; _corporation_pages() globs wiki/corporations/*.md (sorted, index.md excluded, fail-closed) into IMPORT_ECONOMICS_SOURCES; paths.py re-exports from the registry. check-systems-db confirms the stamp now covers the 155 pages. | | T2 | No reset pass for corp_specialization/hq_placement | **Fixed.** NULL-reset for all corps before the authored re-apply, mirroring the settlement_class reset-before-override pattern. Removed-key case verified: a corp dropped from the authored set reverts to all-NULL on the next run. | | M1 | somatic-futures miscategorized as professional_services | **Fixed.** New licensed_clinical_services vocabulary value (31st: anchor medical_services, NonPhysical, Specialist, override concentrated — passes V-SES-03), corp_hq_placement.toml stanza (CityTenant; no standalone_economic_role per the file's Standalone-only convention), somatic-futures retagged. Placement unchanged (CityTenant at Kallast, GJ144d). 155/155 specialized; vocabulary 31 values / placement map 31 stanzas. | Gate re-run on push: fmt/clippy/full cargo test/deny/ruff/test-tooling/stamp all green. Reviewers have been asked to verify their own findings for the re-review verdict.
jpmschweitzer commented 2026-07-16 13:08:36 +02:00
Author
Owner
Copy Link
Copy Source

Re-review verdict: APPROVED

All three reviewers re-verified their own findings against branch tip 9bb533a8b — each independently confirmed the fixes in code and against the committed DB, not just from the fix-round claims:

  • Hoshe: APPROVE. H1 reset+derive confirmed (corporations.py:373-382); GJ702B corps verified at GJ702Bb; dry-run reproduces "154 resolved (recomputed every run)". H2 city-bearing tier confirmed (atlas.py:228-235) — re-ran the original 9-system divergence check: all 8 belt-settlement systems resolve to the belt, both live tenancies preserved. H3 governance wording now accurate. No new issues.
  • Tyre: APPROVE. T1 verified end-to-end: corp-page frontmatter edits now change the stamped SHA. T2 poison-tested: removed key reverts all three columns to NULL. H2 hysteresis ruled architecturally acceptable: committed DB is a self-consistent fixpoint (0 changes on re-run), resolution is order-independent, failure modes bounded and loud; residual observation (committed DB mildly load-bearing for 11 belt-tenant tiebreaks) disclosed and traced to the T-1115 content gap — not a blocker.
  • Miri: APPROVE. licensed_clinical_services style-consistent and correctly distinct from longevity_monopoly on the market-structure axis; placement stanza well-formed; boundary scan (sato-medical, salud-alliance, calluna-wellness, prometheus-labs, spitsbergen-retreat) confirms no other corp moves — the line is Assembly medical licensure of the clinical practice itself.

Merging to main and closing.

## Re-review verdict: APPROVED All three reviewers re-verified their own findings against branch tip 9bb533a8b — each independently confirmed the fixes in code and against the committed DB, not just from the fix-round claims: - **Hoshe: APPROVE.** H1 reset+derive confirmed (corporations.py:373-382); GJ702B corps verified at GJ702Bb; dry-run reproduces "154 resolved (recomputed every run)". H2 city-bearing tier confirmed (atlas.py:228-235) — re-ran the original 9-system divergence check: all 8 belt-settlement systems resolve to the belt, both live tenancies preserved. H3 governance wording now accurate. No new issues. - **Tyre: APPROVE.** T1 verified end-to-end: corp-page frontmatter edits now change the stamped SHA. T2 poison-tested: removed key reverts all three columns to NULL. H2 hysteresis ruled architecturally acceptable: committed DB is a self-consistent fixpoint (0 changes on re-run), resolution is order-independent, failure modes bounded and loud; residual observation (committed DB mildly load-bearing for 11 belt-tenant tiebreaks) disclosed and traced to the T-1115 content gap — not a blocker. - **Miri: APPROVE.** licensed_clinical_services style-consistent and correctly distinct from longevity_monopoly on the market-structure axis; placement stanza well-formed; boundary scan (sato-medical, salud-alliance, calluna-wellness, prometheus-labs, spitsbergen-retreat) confirms no other corp moves — the line is Assembly medical licensure of the clinical practice itself. Merging to main and closing.
jpmschweitzer closed this pull request 2026-07-16 13:12:26 +02:00

Pull request closed

This pull request cannot be reopened because the branch was deleted.
Sign in to join this conversation.
Reviewers
No Reviewers
Labels
No items
No labels
Milestone
No items
No Milestone
Projects
Clear projects
No projects
Assignees
Clear assignees
jpmschweitzer (Jeroen Schweitzer)
No Assignees
1 Participants
Notifications
Due Date
No due date set.
Dependencies

No dependencies set.

Reference: jpmschweitzer/settled-reach#177