Lane E R23 (v5.0-alpha): the atlas validator + toowoomba retired (ledger #4)

1. validate_atlas (the G2a format gate). pipeline/validate_atlas.py, wired into validate_manifest so it
   runs in qa gate 3. Judges Lane G's tier-1 atlases from the COMMITTED FILES ALONE -- no dealgod DB, no
   network. No atlases yet => clean pass, so it's in place before G's lands. Enforces: provenance
   {license, attribution, generator, snapshot} (charter law #3 + determinism); NO private-individual item
   fields (law #3); 1 atlas/shop hard cap 2 (C §7.1); <=1024², 2048² only for a 60+ item shop (C §7.2);
   declared atlas_px == the atlas's REAL pixels + every listed atlas exists; schema/UV/bands/dup-ids
   (C §7.3). The licence string is PRINTED, not pattern-matched -- clearing photos for public release is
   a human call, so the gate surfaces it. Dependency-free PNG/WebP header parser. Tested against a
   10-case matrix (real 2048² WebP + 1024² PNG): good passes clean, all 9 violations caught.

2. toowoomba RETIRED (my call, counted). Tried the re-bbox FIRST: 3.0km CBD-wide -> 1.6km on the Ruthven
   St heart. It fixed the named metric (median NN 395.7 -> 89.4m raw, 'passes') but made the real thing
   WORSE: hub density 3/12 -> 2/12 within 160m, the pack's worst (D's alive-darwin is 7/12). Its shops
   are PAIRS strung along a road, not a high street; densest 300m cluster is 3 shops (< MIN_TOWN_SHOPS),
   so no tighter box exists. Shipping it would have gamed the number while leaving a town nobody shops in
   (D: 0 finds / 1417 checks). Config entry kept commented with the rationale. Pack: 21 real + 1 godverse
   = 22 rostered, 1180 shops, all 8 states/territories (QLD keeps westend). selfcheck ALL GREEN 152015.

-> Lane A, before you freeze the spacing warn (#5): (a) your validator sees the CACHE (raw lat/lon) but
   D's numbers are SEATED (post-lift) -- darwin is 27m seated vs 68.3m RAW, so a threshold picked off D's
   table would FALSE-WARN darwin; calibrate on the raw spread (daylesford 11.6 ... braddon 119.2,
   ballarat 254.6 -- a ~200m line has daylight). (b) median NN alone is GAMEABLE: my re-bbox improved it
   4.4x while the town got worse -- the hub fraction is the discriminator, and my raw hub metric
   reproduces D's darwin exactly (7/12), so it's directly encodable (alive ~38-75%, dead <=20%).
   (c) ballarat is a SECOND spread town D's two-town comparison missed (NN 254.6m, hub 4/20 = 20%) -- it
   will trip your warn, correctly. Not retired: the ledger named toowoomba only.

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>

docs/LANES/LANE_E_NOTES.md

@@ -1,5 +1,59 @@
 # LANE E — cross-lane notes
 
+## Round 23 — v5.0-alpha: the atlas validator + the toowoomba call ⟨v5.0-alpha⟩
+
+**1. `validate_atlas` — the G2a format gate (ledger #4).** `pipeline/validate_atlas.py`, wired into
+`validate_manifest` so it runs in qa gate 3. It judges G's tier-1 atlases **from the committed files
+alone — no dealgod DB, no network** (that's the point: the gate must run on m3). No atlases yet ⇒ clean
+pass, so it's in place *before* G's lands. **→ Lane G: this is the format you build to.** It enforces:
+
+| rule | source |
+|---|---|
+| provenance `{license, attribution, generator, snapshot}` present | charter law #3 ("every atlas carries provenance") + `snapshot` pins determinism |
+| **no private-individual fields** on items (`seller`/`owner`/`email`/`phone`/…) | charter law #3 ("no private-individual data, ever") |
+| 1 atlas/shop, **hard cap 2** | LANE_C_PUB §7.1 |
+| **≤1024²**; 2048² only for a **60+ item** shop | LANE_C_PUB §7.2 |
+| declared `atlas_px` == the atlas's **real** pixels; every listed atlas exists | index↔atlas consistency |
+| schema + UV in range/not inverted, bands, dup ids, empty title/artist, atlas∈atlases[] | §7.3 (schema confirmed verbatim) |
+
+The licence string is **printed** by the gate, not pattern-matched — clearing photos for public release is
+a human call (charter law #3), so the gate surfaces it rather than pretending to rule on it. **Tested
+against a 10-case matrix** (real 2048² WebP + 1024² PNG headers, dependency-free parser): good passes
+clean; all 9 violations caught. A gate that's never caught anything isn't a gate.
+
+**2. Toowoomba — RETIRED (my call, counted). I tried the re-bbox first; the data said retire.**
+D's finding: DEAD — 0 patronage finds in 1417 checks. I re-bboxed 3.0 km CBD-wide → 1.6 km onto the
+Ruthven St heart. Result:
+
+| | median NN (the named metric) | hub ≤160 m (D's real test) |
+|---|---|---|
+| toowoomba @3.0 km | 395.7 m ❌ | 3/12 |
+| **toowoomba @1.6 km (re-bbox)** | **89.4 m ✅ "passes"** | **2/12 (17%) — worst in the pack** |
+| darwin (D: ALIVE) | 68.3 m | 7/12 (58%) |
+
+**The re-bbox fixed the metric and made the town worse.** Toowoomba's shops are *pairs* strung along a
+road, not a high street — close-in-pairs flatters median NN while no hub exists. Its densest 300 m
+cluster is **3 shops** (below `MIN_TOWN_SHOPS`), so no tighter box exists either. Shipping it would have
+gamed the number while leaving a town nobody shops in. **Retired** (config entry kept commented with the
+rationale so nobody re-adds it blindly). Pack: **21 real + 1 godverse = 22 rostered · 1180 shops · all 8
+states/territories** (QLD keeps westend, 11 heroes/44 shops). selfcheck **ALL GREEN 152015/152015**.
+
+### ⚠️ → Lane A: two things before you freeze the spacing warn (#5)
+1. **Your validator sees the CACHE (raw lat/lon); D's numbers are SEATED (post-lift).** They differ:
+   **darwin — D 27 m seated vs 68.3 m raw**; toowoomba 388 seated vs 395.7 raw (spread towns barely move,
+   clustered ones move a lot). **Calibrate off the raw spread, not D's table** — a threshold picked near
+   "darwin 27 = alive" would **false-warn darwin at 68.3**. Raw spread now: daylesford 11.6 · katoomba
+   22.4 · bowral 27.9 · … · launceston 76.8 · braddon 119.2 · **ballarat 254.6**. A ~200 m line has
+   daylight on both sides.
+2. **Median NN alone is gameable — the hub fraction is the discriminator.** My re-bbox is the proof: NN
+   improved 4.4× while the town got worse. Consider warning on **hub density** too (shops within 160 m of
+   the densest shop). **My raw hub metric reproduces D's darwin exactly (7/12)**, so it's directly
+   encodable: alive ≈ 38–75% (castlemaine/darwin/launceston/daylesford), dead ≈ ≤20% (ballarat 20%,
+   retired-toowoomba 17%).
+3. **ballarat is a second spread town** D's two-town comparison never covered: **NN 254.6 m, hub 4/20
+   (20%)** — it *will* trip your warn (correctly). Not retired: the ledger named toowoomba only, and a
+   warn is A's to raise, curation is mine to act on. Flagging for Fable's next-round call.
+
 ## Round 22 — v4.0 THE EPOCH CLOSE: the tri diet (mine) · notes · provenance · the G2a review ⟨v4.0⟩
 
 ### ⚠️ Ledger #1b — the `venue_night` tri diet: **MY four furniture GLBs. Fixed. → F: RE-MEASURE.**

pipeline/build_towns.py

@@ -54,7 +54,16 @@ TOWNS = {
     "daylesford_real":   {"town": "Daylesford",   "state": "VIC", "center": (-37.3430, 144.1430), "span_km": 2.0},
     "geelong_real":      {"town": "Geelong",      "state": "VIC", "center": (-38.1490, 144.3600), "span_km": 3.0},
     "westend_real":      {"town": "West End",     "state": "QLD", "center": (-27.4830, 153.0100), "span_km": 2.2},
-    "toowoomba_real":    {"town": "Toowoomba",    "state": "QLD", "center": (-27.5600, 151.9540), "span_km": 3.0},
+    # ── RETIRED R23 (E's curation call, counted) — toowoomba_real ────────────────────────────────
+    # D's R22 finding: DEAD — median NN 388 m, 3/12 shops near the hub, **0 patronage finds in 1417
+    # checks**. I tried the re-bbox first (3.0 km CBD-wide → 1.6 km on the Ruthven St heart, centre
+    # -27.5610,151.9535): it FIXED the named metric (median NN 395.7 → 89.4 m raw — normal for the
+    # pack) but made the real thing WORSE — hub density 3/12 → **2/12 within 160 m, the worst in the
+    # pack** (D's alive-darwin is 7/12). Toowoomba's shops are PAIRS strung along a road, not a high
+    # street; its densest 300 m cluster is only 3 shops, below MIN_TOWN_SHOPS, so no tighter box
+    # exists. Shipping it would have gamed the spacing metric while leaving a town nobody shops in.
+    # Retired rather than padded. (QLD keeps westend_real, 11 heroes / 44 shops.)
+    # "toowoomba_real":  {"town": "Toowoomba", "state": "QLD", "center": (-27.5600, 151.9540), "span_km": 3.0},
     "northbridge_real":  {"town": "Northbridge",  "state": "WA",  "center": (-31.9490, 115.8570), "span_km": 2.4},
     "adelaide_real":     {"town": "Adelaide",     "state": "SA",  "center": (-34.9240, 138.6050), "span_km": 2.8},
     "hobart_real":       {"town": "Hobart",       "state": "TAS", "center": (-42.8820, 147.3270), "span_km": 2.8},

pipeline/validate_atlas.py

@@ -0,0 +1,172 @@
+#!/usr/bin/env python3
+"""Validate per-shop real-stock atlases — the G2a format gate (v5.0-alpha, ROUND23 ledger #4).
+
+Lane G bakes one atlas + index per real shop from dealgod inventory (tier 1 of the charter's stock
+ladder). This gate is the FORMAT side: it proves G's output is well-formed, inside C's ceilings, and
+carries its provenance — **on m3, from the files alone, with no dealgod DB and no network**. That's the
+point: the gate must run anywhere, on the committed artefacts.
+
+    python3 pipeline/validate_atlas.py        # exit 0 = every atlas valid (or none yet), 1 = a bad one
+
+Contract (published R22 → LANE_E_NOTES E's half · LANE_C_PUB §7 C's ceilings · charter law #3 licence):
+  index  `web/assets/stock_godverse/**/stock_shop_<godverseShopId>_index.json`
+         {version, atlas_px, cell, atlases[], items:[{id,title,artist,price,price_band,atlas,uv}]}
+         + provenance: {license, attribution, generator, snapshot}
+  C's ceilings — 1 atlas/shop (hard cap 2); ≤1024² typical, 2048² only for a 60+ item shop.
+  charter law #3 — every atlas carries provenance; no private-individual data, ever.
+  determinism — `snapshot` pins the inventory query result (G snapshots it like E snapshots Overpass raw).
+
+Fail-soft is NOT this gate's business: a missing atlas is legal at runtime (parody canvas, R7 law). This
+gate only judges atlases that EXIST — a malformed one must fail loudly rather than render garbled sleeves.
+"""
+import json, os, sys, glob, struct
+
+ROOT = os.path.dirname(os.path.dirname(os.path.abspath(__file__)))
+STOCK_G = os.path.join(ROOT, "web", "assets", "stock_godverse")      # Lane G's namespace (charter law #5)
+BANDS = {"bargain", "standard", "collector", "grail"}
+REQUIRED_ITEM = ("id", "title", "artist", "price", "price_band", "atlas", "uv")
+REQUIRED_PROV = ("license", "attribution", "generator", "snapshot")   # charter law #3 + determinism
+MAX_ATLASES = 2          # C: 1 atlas/shop, hard cap 2 (each extra = +1 draw)
+PX_TYPICAL, PX_MAX = 1024, 2048
+BIG_SHOP_ITEMS = 60      # C: 2048² only for a 60+ item shop (else spill to the type-pack)
+# charter law #3: "No private-individual data, ever" — a real shop's stock row must never carry a person.
+FORBIDDEN_ITEM_KEYS = {"seller", "seller_name", "owner", "owner_name", "customer", "user", "username",
+                       "email", "phone", "mobile", "address", "member", "account", "consignor"}
+
+
+def image_size(path):
+    """(w, h) for PNG/WebP from the header alone — dependency-free, house style (cf. glb_stat.py)."""
+    with open(path, "rb") as f:
+        head = f.read(32)
+    if head[:8] == b"\x89PNG\r\n\x1a\n":
+        w, h = struct.unpack(">II", head[16:24])
+        return w, h
+    if head[:4] == b"RIFF" and head[8:12] == b"WEBP":
+        fmt = head[12:16]
+        if fmt == b"VP8X":                                  # extended: 24-bit minus-one
+            b = head[24:30]
+            return (b[0] | b[1] << 8 | b[2] << 16) + 1, (b[3] | b[4] << 8 | b[5] << 16) + 1
+        if fmt == b"VP8 ":                                  # lossy: 14-bit each
+            w, h = struct.unpack("<HH", head[26:30])
+            return w & 0x3FFF, h & 0x3FFF
+        if fmt == b"VP8L":                                  # lossless: packed 14-bit
+            b = struct.unpack("<I", head[21:25])[0]
+            return (b & 0x3FFF) + 1, ((b >> 14) & 0x3FFF) + 1
+    raise ValueError("unrecognised image header (want PNG or WebP)")
+
+
+def validate_index(path):
+    errs, warns = [], []
+    name = os.path.basename(path)
+    try:
+        idx = json.load(open(path))
+    except Exception as e:
+        return [f"{name}: does not parse: {e}"], []
+    d = os.path.dirname(path)
+
+    # ── provenance (charter law #3) — every atlas carries it, and the gate PRINTS the licence ──
+    miss = [f for f in REQUIRED_PROV if not str(idx.get(f, "")).strip()]
+    if miss:
+        errs.append(f"{name}: missing/empty provenance {miss} (charter law #3 — every atlas carries provenance)")
+
+    # ── C's draw ceiling (LANE_C_PUB §7.1) ──
+    atlases = idx.get("atlases", [])
+    if not isinstance(atlases, list) or not atlases:
+        errs.append(f"{name}: no atlases[] listed")
+        atlases = []
+    if len(atlases) > MAX_ATLASES:
+        errs.append(f"{name}: {len(atlases)} atlases — C's hard cap is {MAX_ATLASES} "
+                    f"(1/shop; a 60+ item shop spills to the shared type-pack instead of fragmenting)")
+    elif len(atlases) == 2:
+        warns.append(f"{name}: 2 atlases (+1 draw) — C's rule is 1/shop; 2 is the cap, not the target")
+
+    items = idx.get("items", [])
+    if not isinstance(items, list) or not items:
+        errs.append(f"{name}: no items")
+        items = []
+
+    # ── C's texture/VRAM ceiling (§7.2) + index↔atlas consistency (the declared size must be REAL) ──
+    declared = idx.get("atlas_px")
+    for a in atlases:
+        ap = os.path.join(d, a)
+        if not os.path.isfile(ap):
+            errs.append(f"{name}: atlas {a!r} listed but missing on disk")
+            continue
+        try:
+            w, h = image_size(ap)
+        except Exception as e:
+            errs.append(f"{name}: atlas {a!r} unreadable: {e}")
+            continue
+        if declared and (w != declared or h != declared):
+            errs.append(f"{name}: atlas {a!r} is {w}×{h} but index declares atlas_px={declared}")
+        big = len(items) >= BIG_SHOP_ITEMS
+        cap = PX_MAX if big else PX_TYPICAL
+        if max(w, h) > cap:
+            errs.append(f"{name}: atlas {a!r} {w}×{h} exceeds C's ceiling {cap}² "
+                        f"({'60+ item shop' if big else f'{len(items)}-item shop → ≤{PX_TYPICAL}²'})")
+        if max(w, h) > PX_TYPICAL and not big:
+            warns.append(f"{name}: atlas {a!r} is {w}×{h} for {len(items)} items — prefer ≤{PX_TYPICAL}²")
+
+    # ── the index itself (schema confirmed verbatim in LANE_C_PUB §7.3) ──
+    ids = set()
+    for i, it in enumerate(items):
+        if not isinstance(it, dict):
+            errs.append(f"{name}[{i}]: item is not an object"); continue
+        m = [f for f in REQUIRED_ITEM if f not in it]
+        if m:
+            errs.append(f"{name}[{i}]: missing {m}"); continue
+        priv = sorted(set(k.lower() for k in it) & FORBIDDEN_ITEM_KEYS)
+        if priv:
+            errs.append(f"{name}[{it['id']}]: private-individual field(s) {priv} — charter law #3 forbids it")
+        if not str(it["title"]).strip() or not str(it["artist"]).strip():
+            errs.append(f"{name}[{it['id']}]: empty title/artist")
+        if it["id"] in ids:
+            errs.append(f"{name}: duplicate id {it['id']!r}")
+        ids.add(it["id"])
+        if it["price_band"] not in BANDS:
+            errs.append(f"{name}[{it['id']}]: bad price_band {it['price_band']!r}")
+        if not isinstance(it["price"], (int, float)) or it["price"] < 0:
+            errs.append(f"{name}[{it['id']}]: bad price {it['price']!r}")
+        if it["atlas"] not in atlases:
+            errs.append(f"{name}[{it['id']}]: atlas {it['atlas']!r} not in atlases[]")
+        uv = it["uv"]
+        if not (isinstance(uv, list) and len(uv) == 4 and all(isinstance(v, (int, float)) for v in uv)):
+            errs.append(f"{name}[{it['id']}]: uv must be [u0,v0,u1,v1] numbers"); continue
+        u0, v0, u1, v1 = uv
+        if not (0 <= u0 < u1 <= 1 and 0 <= v0 < v1 <= 1):     # UV origin top-left; C flips V (R7 law)
+            errs.append(f"{name}[{it['id']}]: uv out of range / inverted: {uv}")
+    return errs, warns
+
+
+def main():
+    idxs = sorted(glob.glob(os.path.join(STOCK_G, "**", "stock_shop_*_index.json"), recursive=True))
+    if not idxs:
+        print("atlas-QA: no per-shop atlases yet (Lane G's G2a) — nothing to validate")
+        return 0
+    all_errs, all_warns = [], []
+    for p in idxs:
+        e, w = validate_index(p)
+        all_errs += e; all_warns += w
+        idx = {}
+        try:
+            idx = json.load(open(p))
+        except Exception:
+            pass
+        print(f"  {os.path.basename(p):40} items={len(idx.get('items', [])):>3} "
+              f"atlases={len(idx.get('atlases', []))} px={idx.get('atlas_px')} "
+              f"snapshot={str(idx.get('snapshot', '—'))[:24]} {'OK' if not e else 'ERR(' + str(len(e)) + ')'}")
+        if idx.get("license"):                    # licence law: the gate SHOWS it, a human clears it
+            print(f"      licence: {idx['license']}")
+    for w in all_warns:
+        print(f"  WARN {w}")
+    for e in all_errs:
+        print(f"  ERR  {e}")
+    if all_errs:
+        print(f"atlas-QA FAIL — {len(all_errs)} error(s), {len(all_warns)} warning(s)")
+        return 1
+    print(f"atlas-QA OK — {len(idxs)} per-shop atlas(es) valid, {len(all_warns)} warning(s)")
+    return 0
+
+
+if __name__ == "__main__":
+    sys.exit(main())

pipeline/validate_manifest.py

@@ -152,6 +152,15 @@ def main():
     except Exception as e:
         warn(f"pack-QA skipped: {e}")
 
+    # per-shop atlas QA (v5 G2a, ROUND23 E #4): Lane G's tier-1 atlases are gate-checked from the
+    # committed files alone — no dealgod DB, no network. No atlases yet ⇒ clean pass.
+    try:
+        import validate_atlas
+        if validate_atlas.main() != 0:
+            err("per-shop atlas validation failed (see atlas-QA errors above)")
+    except Exception as e:
+        warn(f"atlas-QA skipped: {e}")
+
     print(f"manifest v{m.get('version')} — facades {len(facade)}, "
           f"skins {sum(len(v) if isinstance(v, list) else (len(v) if isinstance(v, dict) else 0) for v in sk.values())} groups, "
           f"GLBs {n_glb}, audio {n_audio}")

web/assets/towns/SOURCES.md

@@ -20,20 +20,26 @@ the "© OpenStreetMap contributors (ODbL)" credit visible.
 **`index.json` is the machine-readable roster** (key/town/state/shops/roads) — read it rather than
 hardcoding a list. Every cache carries the same terms; per-town numbers live in each cache's `counts`.
 
-**The v4.0 pack (R21): 22 towns · 1192 shops · 298 heroes**, every one fetched from OSM Overpass with a
-bounded per-town bbox, ODbL 1.0, raw snapshots in `_raw/` (shops + roads per town), fetched 2026-07-16.
+**The pack: 21 real towns · 1180 shops · 295 heroes** (+ Lane G's `newtown_godverse` = 22 rostered in
+`index.json`), every one fetched from OSM Overpass with a bounded per-town bbox, ODbL 1.0, raw snapshots
+in `_raw/` (shops + roads per town), fetched 2026-07-16.
 
 | state | towns |
 |---|---|
 | NSW (6) | katoomba · newtown · glebe · marrickville · newcastle (Hamilton strip) · bowral |
 | VIC (7) | fitzroy · brunswick · geelong · bendigo · ballarat · castlemaine · daylesford |
 | WA (2) | fremantle · northbridge |
-| QLD (2) | westend · toowoomba |
+| QLD (1) | westend |
 | TAS (2) | hobart · launceston |
 | SA (1) | adelaide |
 | ACT (1) | braddon |
 | NT (1) | darwin |
 
+**Retired (R23, counted):** `toowoomba_real` — D's clustering audit measured it DEAD (0 patronage finds
+in 1417 checks); a re-bbox onto the Ruthven St heart fixed the spacing metric (median NN 396 → 89 m) but
+made the hub worse (2/12 within 160 m, the pack's worst), and its densest 300 m cluster is only 3 shops.
+Cache + raw snapshots removed; the config entry stays commented with the rationale. See LANE_E_NOTES R23.
+
 ## The density widening (v4.0-beta, ROUND20 ledger #2)
 
 The fetch covers the **secondhand heroes** plus the everyday main-street **texture** (bakery/cafe/deli/

web/assets/towns/index.json

@@ -172,14 +172,6 @@
    "roads": true,
    "source": "osm"
   },
-  {
-   "key": "toowoomba_real",
-   "town": "Toowoomba",
-   "state": "QLD",
-   "shops": 12,
-   "roads": true,
-   "source": "osm"
-  },
   {
    "key": "westend_real",
    "town": "West End",