fix(simulation): one coastline at every rung + feature-name pipeline + legacy field retirement (T-1160, T-1169, T-1159) #208

2026-07-25T16:58:06+02:00

jpmschweitzer commented

2026-07-25 16:58:06 +02:00

Summary

The server trio of the Atlas polish batch.

T-1160 (high) — the orbital rung applies coast_warp_px. Global/Region sampled the heightmap ocean mask raw while every finer rung samples at the coast-warped position — a structurally different coastline at the seam. Audited on all 267 real bodies at region spacing (coastal band): 5.01% land/ocean disagreement, mean displacement ~6.0 km, max ~18.9 km. Fork taken: APPLY the warp — derive_orbital_at_metres now runs invent_primitives' steps 1-3 before sampling, still skipping detail-scatter (its 32.8 km octave ceiling is below region spacing; the warp's octaves reach 262 km, which is why skipping the warp was wrong). Cost measured ~784 ns/cell (~15 ms on an Earth-class Global canvas). The ignored audit test preserves the pre-fix baseline. Golden re-pinned (region rows only; district/quarter byte-identical). project.yaml → 0.4.3. Araminta's sign-off on the fork is required by the ticket — she has the review seat.

T-1169 (low) — feature-name pipeline wired, three dormant pieces connected. atlas_feature_names populated at regen (17,891 rows: 15,190 mountain + 2,701 river; new generator step registered in the T-1067 registry, systems.db regenerated, stamp fresh); attach_feature_names wired into the cascade's Topography block with pools threaded DB-free through AnalyzeBody (D-225), assignments stored on Layer1Output/BodyWorldState; FeatureNamesRequest/Response read proxy as the bridge's 7th tagged envelope (D-236). Client label draw is deliberately absent: implementation proved both river and mountain labels need a wire-carried position (the pool is position-free; course polylines aren't correlated with named attractors by construction) — the single design pass is T-1195; scope adjudication on both tickets.

T-1159 (low) — legacy window_granularity u32 retired. The u32 request field, response echo, u32::MAX sentinel, and legacy resolver are gone server-side; client encode paths and the caller-less atlas_window_cache legacy key dropped; fixtures regenerated; the T-1150 aliasing regression test drives through the surviving enum. The district_window carrier survives byte-compatible per D-255(c).

Verification

Triple gate green on push: client 3,554/3,554, server 1,954 lib + bridge_tcp 22 + all golden/believability/acceptance suites, clippy/deny, tooling tests, systems.db stamp check live with the new generator source. Targeted client suites for the T-1159 changes verified pre-gate (protocol 72, window-cache 30, data-delivery 60). Merge seam with PR #207 clean (this branch never touches the annotation layer — the label-draw fold made the collision structurally impossible). Reviews next; the eyeball (coastline-consistency check at the Global→Region→District seam) runs on the final tip — merge waits on it.

🤖 Generated with Claude Code

## Summary The server trio of the Atlas polish batch. **T-1160 (high) — the orbital rung applies coast_warp_px.** Global/Region sampled the heightmap ocean mask raw while every finer rung samples at the coast-warped position — a structurally different coastline at the seam. Audited on all 267 real bodies at region spacing (coastal band): 5.01% land/ocean disagreement, mean displacement ~6.0 km, max ~18.9 km. Fork taken: APPLY the warp — `derive_orbital_at_metres` now runs invent_primitives' steps 1-3 before sampling, still skipping detail-scatter (its 32.8 km octave ceiling is below region spacing; the warp's octaves reach 262 km, which is why skipping the warp was wrong). Cost measured ~784 ns/cell (~15 ms on an Earth-class Global canvas). The ignored audit test preserves the pre-fix baseline. Golden re-pinned (region rows only; district/quarter byte-identical). project.yaml → 0.4.3. **Araminta's sign-off on the fork is required by the ticket — she has the review seat.** **T-1169 (low) — feature-name pipeline wired, three dormant pieces connected.** atlas_feature_names populated at regen (17,891 rows: 15,190 mountain + 2,701 river; new generator step registered in the T-1067 registry, systems.db regenerated, stamp fresh); attach_feature_names wired into the cascade's Topography block with pools threaded DB-free through AnalyzeBody (D-225), assignments stored on Layer1Output/BodyWorldState; FeatureNamesRequest/Response read proxy as the bridge's 7th tagged envelope (D-236). **Client label draw is deliberately absent:** implementation proved both river and mountain labels need a wire-carried position (the pool is position-free; course polylines aren't correlated with named attractors by construction) — the single design pass is T-1195; scope adjudication on both tickets. **T-1159 (low) — legacy window_granularity u32 retired.** The u32 request field, response echo, u32::MAX sentinel, and legacy resolver are gone server-side; client encode paths and the caller-less atlas_window_cache legacy key dropped; fixtures regenerated; the T-1150 aliasing regression test drives through the surviving enum. The district_window carrier survives byte-compatible per D-255(c). ## Verification Triple gate green on push: client 3,554/3,554, server 1,954 lib + bridge_tcp 22 + all golden/believability/acceptance suites, clippy/deny, tooling tests, systems.db stamp check live with the new generator source. Targeted client suites for the T-1159 changes verified pre-gate (protocol 72, window-cache 30, data-delivery 60). Merge seam with PR #207 clean (this branch never touches the annotation layer — the label-draw fold made the collision structurally impossible). Reviews next; the eyeball (coastline-consistency check at the Global→Region→District seam) runs on the final tip — merge waits on it. 🤖 Generated with [Claude Code](https://claude.com/claude-code)

jpmschweitzer added 5 commits 2026-07-25 16:58:06 +02:00

fix(simulation): orbital rung applies coast_warp_px — one coastline at every rung (T-1160) befdb689c1

The Global/Region rungs sampled the heightmap ocean mask raw while every
finer rung samples it at the coast-warped position — a structurally
different coastline at the orbital-to-district seam. Audited on all 267
real body heightmaps at region spacing, coastal band only: 5.01%
land/ocean classification disagreement (2,911 of 58,073 cells), mean
displacement ~6.0 km, max ~18.9 km. Fork taken: APPLY the warp —
derive_orbital_at_metres now runs invent_primitives' steps 1-3 (driver
climate -> coast character -> coast_warp_px) before sampling, still
skipping detail-scatter (its octave ceiling, 32.8 km, is below region
spacing; the WARP's octaves reach 262 km, which is why skipping it was
wrong). Cost measured: ~784 ns/cell added (~15 ms on an Earth-class
Global canvas; low hundreds of ms parallel at the 8.3M-cell ceiling).
The ignored audit test preserves the pre-fix baseline for the record.
Golden re-pinned — region rows only, district/quarter byte-identical.
project.yaml 0.4.2 -> 0.4.3: orbital canvas bytes changed, client disk
caches must miss. Awaiting Araminta's review-seat sign-off per ticket.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>

feat(simulation): feature-name pipeline wired + legacy window_granularity u32 retired (T-1169, T-1159) 8da9670e0f

One commit for two tickets whose changes share the bridge/plugin
plumbing files. T-1169 connects the three dormant feature-name pieces:
atlas_feature_names populated at regen (17,891 rows — 15,190 mountain,
2,701 river — via populate_atlas_feature_names mirroring the city-names
importer; systems.db regenerated, stamp fresh), attach_feature_names
wired into the cascade's Topography block with name pools threaded
DB-free through AnalyzeBody (D-225 pattern) and assignments stored on
Layer1Output/BodyWorldState for future consumers, and a
FeatureNamesRequest/Response read proxy as the bridge's 7th tagged
envelope (D-236 pattern, both SimBridge impls). Client label DRAW is
deliberately NOT here — implementation proved both river and mountain
labels need a wire-carried position (the pool is position-free; course
polylines aren't correlated with the named attractors by construction) —
deferred to T-1195's single design pass. cascade_layer1 golden re-pinned
(additive feature_names field).

T-1159 retires the legacy u32 granularity field fully shadowed by
window_granularity_v2: AtlasLayerRequest.window_granularity,
DistrictWindowLayer.granularity echo, the u32::MAX sentinel, and
resolve_window_granularity are gone server-side; client encode paths and
the caller-less atlas_window_cache legacy key component dropped;
msgpack fixtures regenerated; the T-1150 aliasing regression test now
drives through the surviving enum field. The district_window carrier
itself survives byte-compatible per D-255(c).

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>

chore(meta): update changelog 1eb596ec78

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>

Merge remote-tracking branch 'origin/main' into atlas-feature-names 497f9dc94d

# Conflicts:
#	CHANGELOG.md

style(simulation): gate bounce — cargo fmt 71f70f035e

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>

jpmschweitzer referenced this issue from a commit

2026-07-25 17:05:57 +02:00

chore(meta): pql changelog — T-1194/T-1195 filed, polish batch transitions

jpmschweitzer referenced this issue from a commit

2026-07-25 17:22:58 +02:00

fix(simulation): one shared invent_coastal_position — PR #208 review round (T-1160)

jpmschweitzer added 1 commit 2026-07-25 17:22:58 +02:00

fix(simulation): one shared invent_coastal_position — PR #208 review round (T-1160) 077d787c4a

Hoshe found the concrete residual: the hand-copied driver block called
derive_temperature_c unconditionally where invent_primitives prefers the
lapse-adjusted region baseline on non-airless bodies — driver_temp feeds
glaciation/moisture into the warp magnitude, so orbital displacement
could still differ from District at the same position (the ticket's
defect class, one step upstream); the copy structurally couldn't branch
right because region_baseline_at_district was computed after it. Tyre
demanded the structural cure: steps 1-3 now live in ONE shared helper
(invent_coastal_position -> warped position + CoastCharacter);
invent_primitives composes helper + detail-scatter; the orbital path
hoists the baseline and makes a single helper call. The audit test's own
'District-style' side turned out to be a THIRD copy carrying the same
bug — rewired to the real helper (raw pre-fix side untouched as the
historical baseline; refreshed: 5.01% disagreement, mean 5,957.9 m,
748.9 ns/cell). Golden verified byte-identical under forced regen, with
the reason traced: the wire temperature never flowed through the buggy
internal branch, and the fixture's probe positions cross no discrete
boundary — the 267-body audit is the instrument that sees the drift.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>

jpmschweitzer commented

2026-07-25 17:23:33 +02:00

Review: atlas-feature-names -> main (type: code)

Reviewers: Hoshe (code quality) + Tyre (architecture) + Araminta (the T-1160 fork sign-off her PR #192 finding entitles her to). All three REQUEST_CHANGES; all findings addressed in 077d787c4 or resolved as bookkeeping.

Hoshe: REQUEST_CHANGES

#	File	Issue	Resolution
1	district_profile.rs:2056	The replicated driver block called derive_temperature_c UNCONDITIONALLY, where invent_primitives prefers the lapse-adjusted region baseline on non-airless bodies — driver_temp feeds glaciation/moisture into the warp magnitude, so the orbital rung could still displace differently than District at the same position: T-1160's defect class, one step upstream. The copy structurally couldn't branch right (region_baseline_at_district was computed after it)	Fixed structurally — steps 1-3 extracted into one shared `invent_coastal_position` helper carrying the baseline-preferring branch; the orbital path hoists the baseline and makes a single helper call (`077d787c4`)

Hoshe's verified-clean list: T-1169's parameterized SQL, DB-free worker discipline, demux collision-safety, both SimBridge impls wired; T-1159's zero survivor references; the golden re-pin scope claim verified exactly.

Tyre: REQUEST_CHANGES

#	Issue	Resolution
1	The orbital path was a hand-copied second implementation of steps 1-3 — the exact drift disease D-256 exists to prevent; same-PR refactor demanded	Fixed by the same change — one implementation exists; invent_primitives composes helper + detail-scatter. Bonus: the audit test's own "warped read" side turned out to be a THIRD copy carrying the same bug — rewired to the real helper (the raw pre-fix side stays as the historical baseline)
2	The T-1169/T-1160 scope adjudication references T-1195, which the branch's planning store cannot resolve	Resolved as bookkeeping lag, substance fixed — T-1195 exists (created from main before the review); its write-through changelog rows were awaiting a bookkeeping commit, now committed and pushed to main (`7a55685bf`); resolves on this branch's merge-from-main

Tyre's confirmations: cache-hit==cache-miss still covers the orbital path; no persistent server-side store serves stale canvases; the 7th envelope mirrors existing patterns cleanly; stored name assignments are re-derivable (derive-don't-store holds); the district_window carrier byte-safe for its one consumer.

Araminta (T-1160 sign-off): REQUEST_CHANGES

Endorses the APPLY fork on the audit evidence (she checked the octave-ceiling math and the golden re-pin honesty herself) — but withholds sign-off until the visual verification exists: "the whole point of my PR #192 finding was that this class of bug is invisible in a code diff." Her capture spec (GJ380c + GJ820Bc; Global/Region/District frames on the same coastal stretch; a matched-crop Region/District overlay composite as the actual check; a main-vs-branch before/after pair as the strongest artifact; Quarter-vs-District relief differentiation as a separate ride-along pass on a mountainous crop) is the eyeball brief. Her T-1169 deferral concern: none — "a real design blocker, not a scope-avoidance move."

Fix-round verification

Refreshed audit under the shared helper: 5.01% disagreement (unchanged), mean 5,957.9 m / max 18,867.1 m, 748.9 ns/cell. Golden verified byte-identical under FORCED regen, with the reason traced (the wire temperature never flowed through the buggy internal branch; the fixture's probe positions cross no discrete boundary — the 267-body audit is the instrument that sees the drift). All targeted suites green; full triple gate green on push.

Verdict: CHANGES REQUESTED -> code addressed in `077d787c4`; the eyeball to Araminta's spec runs on this tip, and her re-look on its captures gates the merge.

# Review: atlas-feature-names -> main (type: code) Reviewers: Hoshe (code quality) + Tyre (architecture) + Araminta (the T-1160 fork sign-off her PR #192 finding entitles her to). All three REQUEST_CHANGES; all findings addressed in `077d787c4` or resolved as bookkeeping. ### Hoshe: REQUEST_CHANGES | # | File | Issue | Resolution | |---|------|-------|------------| | 1 | district_profile.rs:2056 | The replicated driver block called derive_temperature_c UNCONDITIONALLY, where invent_primitives prefers the lapse-adjusted region baseline on non-airless bodies — driver_temp feeds glaciation/moisture into the warp magnitude, so the orbital rung could still displace differently than District at the same position: T-1160's defect class, one step upstream. The copy structurally couldn't branch right (region_baseline_at_district was computed after it) | **Fixed structurally** — steps 1-3 extracted into one shared `invent_coastal_position` helper carrying the baseline-preferring branch; the orbital path hoists the baseline and makes a single helper call (`077d787c4`) | Hoshe's verified-clean list: T-1169's parameterized SQL, DB-free worker discipline, demux collision-safety, both SimBridge impls wired; T-1159's zero survivor references; the golden re-pin scope claim verified exactly. ### Tyre: REQUEST_CHANGES | # | Issue | Resolution | |---|-------|------------| | 1 | The orbital path was a hand-copied second implementation of steps 1-3 — the exact drift disease D-256 exists to prevent; same-PR refactor demanded | **Fixed by the same change** — one implementation exists; invent_primitives composes helper + detail-scatter. Bonus: the audit test's own "warped read" side turned out to be a THIRD copy carrying the same bug — rewired to the real helper (the raw pre-fix side stays as the historical baseline) | | 2 | The T-1169/T-1160 scope adjudication references T-1195, which the branch's planning store cannot resolve | **Resolved as bookkeeping lag, substance fixed** — T-1195 exists (created from main before the review); its write-through changelog rows were awaiting a bookkeeping commit, now committed and pushed to main (`7a55685bf`); resolves on this branch's merge-from-main | Tyre's confirmations: cache-hit==cache-miss still covers the orbital path; no persistent server-side store serves stale canvases; the 7th envelope mirrors existing patterns cleanly; stored name assignments are re-derivable (derive-don't-store holds); the district_window carrier byte-safe for its one consumer. ### Araminta (T-1160 sign-off): REQUEST_CHANGES Endorses the APPLY fork on the audit evidence (she checked the octave-ceiling math and the golden re-pin honesty herself) — but withholds sign-off until the visual verification exists: "the whole point of my PR #192 finding was that this class of bug is invisible in a code diff." Her capture spec (GJ380c + GJ820Bc; Global/Region/District frames on the same coastal stretch; a matched-crop Region/District overlay composite as the actual check; a main-vs-branch before/after pair as the strongest artifact; Quarter-vs-District relief differentiation as a separate ride-along pass on a mountainous crop) is the eyeball brief. Her T-1169 deferral concern: none — "a real design blocker, not a scope-avoidance move." ### Fix-round verification Refreshed audit under the shared helper: 5.01% disagreement (unchanged), mean 5,957.9 m / max 18,867.1 m, 748.9 ns/cell. Golden verified byte-identical under FORCED regen, with the reason traced (the wire temperature never flowed through the buggy internal branch; the fixture's probe positions cross no discrete boundary — the 267-body audit is the instrument that sees the drift). All targeted suites green; full triple gate green on push. ### Verdict: CHANGES REQUESTED -> code addressed in `077d787c4`; the eyeball to Araminta's spec runs on this tip, and her re-look on its captures gates the merge.

jpmschweitzer commented

2026-07-25 18:14:57 +02:00

Eyeball to Araminta's spec — executed, and her sign-off: APPROVE (merge gate cleared)

Live runs on 077d787c4: branch client+server vs the MAIN checkout's pre-fix client+server (never mixed — each ran in spawn mode against its own binary), GJ380c + GJ820Bc.

Coastline seam (her "actual check"): the matched-crop Region/District composites show one coastline — headland, bay recess, peninsula all landing at the same position across rungs on both bodies; no jog, no one-rung-only land/ocean strip.

Before/after (her strongest artifact): coastal-band-only pixel diff between main's pre-fix and the branch's post-fix Region render = 5.12% (1564/30545) — independently corroborating the analytic audit's 5.01% by a completely different method. Correct reading confirmed: the ~19 km max displacement is sub-cell at Region's 204.8 km resolution, so the signature is scattered single-cell coastal-transition reclassifications (visible in before-after-zoom.png), not a shifted line — exactly what a correct warp-position fix should produce.

Self-caught harness artifact: the first "before" attempt captured Region cold without Global first, hitting the documented uncapped fallback and producing a wrapped frame that could have been misread as a pre-fix defect — caught, excluded, re-run clean.

Quarter-vs-District relief (separate pass per her instruction): neither spec body has Alpine terrain (both tectonics:low — proven by exhaustive scans). Substituted plains/river micro-texture; Quarter clearly finer than District. Araminta's ruling: option (b) — accepted for this merge; her PR #192 companion note stays open pending a true tectonics:high capture, now tracked as T-1196 (with the body-pool-recording instruction so the next pass doesn't re-scan blind).

Her verdict on the pixels: "This is what 'one coastline, not two' looks like... my PR #192 finding is closed." APPROVE.

Captures: worktree .cache/screenshots/t1160-eyeball/{GJ380c,GJ820Bc}/; pre-fix frames under main's .cache/screenshots/t1160-eyeball/GJ380c/. Process table verified clean. Merging.

# Eyeball to Araminta's spec — executed, and her sign-off: APPROVE (merge gate cleared) Live runs on `077d787c4`: branch client+server vs the MAIN checkout's pre-fix client+server (never mixed — each ran in spawn mode against its own binary), GJ380c + GJ820Bc. **Coastline seam (her "actual check"):** the matched-crop Region/District composites show one coastline — headland, bay recess, peninsula all landing at the same position across rungs on both bodies; no jog, no one-rung-only land/ocean strip. **Before/after (her strongest artifact):** coastal-band-only pixel diff between main's pre-fix and the branch's post-fix Region render = **5.12%** (1564/30545) — independently corroborating the analytic audit's 5.01% by a completely different method. Correct reading confirmed: the ~19 km max displacement is sub-cell at Region's 204.8 km resolution, so the signature is scattered single-cell coastal-transition reclassifications (visible in `before-after-zoom.png`), not a shifted line — exactly what a correct warp-position fix should produce. **Self-caught harness artifact:** the first "before" attempt captured Region cold without Global first, hitting the documented uncapped fallback and producing a wrapped frame that could have been misread as a pre-fix defect — caught, excluded, re-run clean. **Quarter-vs-District relief (separate pass per her instruction):** neither spec body has Alpine terrain (both tectonics:low — proven by exhaustive scans). Substituted plains/river micro-texture; Quarter clearly finer than District. **Araminta's ruling: option (b)** — accepted for this merge; her PR #192 companion note stays open pending a true tectonics:high capture, now tracked as **T-1196** (with the body-pool-recording instruction so the next pass doesn't re-scan blind). **Her verdict on the pixels:** "This is what 'one coastline, not two' looks like... my PR #192 finding is closed." APPROVE. Captures: worktree `.cache/screenshots/t1160-eyeball/{GJ380c,GJ820Bc}/`; pre-fix frames under main's `.cache/screenshots/t1160-eyeball/GJ380c/`. Process table verified clean. Merging.

jpmschweitzer closed this pull request

2026-07-25 18:20:21 +02:00

jpmschweitzer referenced this issue from a commit

2026-07-25 19:05:16 +02:00

chore(meta): pql changelog — batch-4 activation rows (T-948 closed as delivered, T-964/T-971 activation + appends)

jpmschweitzer referenced this pull request

2026-07-25 19:16:41 +02:00

feat(ui): AtlasAgentInterface — observe/act named-intent control channel (D-226, T-971) #209