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

feat(simulation): T-1156 wave 1 — rivers, basins, attractors onto the Atlas zoom ladder #195

Closed
jpmschweitzer wants to merge 0 commits from rivers-ladder into main
pull from: rivers-ladder
merge into: :main
Conversation 1 Commits - Files Changed -
+1931 -3
jpmschweitzer commented 2026-07-23 09:03:03 +02:00
Owner
Copy Link
Copy Source

Summary

T-1156 wave 1: the nature layer's hydrology arrives on the Atlas zoom ladder — rivers, drainage basins, and attractors drawn over the derived terrain, per-rung filtered. Wave 2 (everything civilizational) remains explicitly deferred per the nature-first order.

The design pass (before any code)

Tyre's carrier ruling reframed the ticket: the river skeleton already rides the wire — AtlasLayerResponse.layer1 has serialized RiverNetwork/drainage_basins/attractors whole-body since #952, and the retired draw code survives intact. So wave 1 is a re-hosting plus one additive field, touching neither D-226 ceiling (windowed family stays at one field; whole-body family untouched — layer1 predates that framing). Per-rung refinement is client-side class filtering — deliberately unlike T-1162's server-side cutoff, because a fixed finite graph has nothing to Nyquist-truncate. The ruling establishes the wave-2 pattern for free: discrete map features ride the whole-body overlay family filtered client-side; only continuous per-metre fields ride the windowed per-cell arrays. Captured as a D-226 note on this branch.

Design-pass honesty, ticketed not buried: T-1168 (riparian vegetation response — near_perennial_water is a hardcoded-false stub, so rivers draw over barren-painted banks for now, deliberately not suppressed per Araminta+Tyre), T-1169 (feature-name labels — three dormant pieces, not "cheap"), T-1170 (the skeleton is 512×256 heightmap-resolution, ~76 km/cell — river courses below Region scale need invention, the linear analog of the coast crinkle).

Server (Dudley) — river_class on RiverNetwork

river_class: Vec<u8> (0=stream 1=tributary 2=trunk) parallel to river_cells, #[serde(default)], computed at drainage-analysis time by log-scaled binning between RIVER_THRESHOLD and the river-restricted max accumulation. The restriction matters: grid-wide max includes below-sea-level ocean cells where accumulation peaks past the coastline — anchoring there starves the trunk class exactly on wet large-ocean bodies (caught in lead review of round 1; GJ1c went from zero trunk cells to a healthy 72/18/3 distribution). By construction every body with rivers has trunk cells, pinned by a non-synthetic regression on the committed GJ1c heightmap. cascade_layer1.json regenerated: every pre-existing field byte-identical (python-verified), only river_class added.

Client (Stig) — the nature overlay on the ladder

New AtlasWindowNatureOverlay node: self-connects to the shared atlas_layers_received broadcast (the established one-signal-N-consumers shape), consumes the whole-body skeleton, draws over the terrain composite. Pure tested coordinate chain layer1_pixel → world_m → district → canvas-local with the vertical convention verified against the server source (row 0 = north pole) and revert-verified pole tests.

Araminta's presentation ruling (supersedes the provisional add-as-you-descend mapping, on the resolution evidence): Region/orbital = full skeleton (radii 0.9/1.4/2.2 px by class, confluence 3.5, mouth double-ring), District = trunk-only at 60%/1.6 px with mouths kept at full landmark styling, Quarter = rivers off until T-1170 — isolated 76 km dots there are placement coincidence, not information. Retired palette reused verbatim. Toggles RVR on / BAS off / ATR off.

Live rounds (the eyeball check earning its keep again)

  1. Zoom-scaled marker sizes: all marker radii were raw constants drawn inside the zoom-scaled canvas — sub-pixel-invisible at orbital fit zoom (~0.014 effective px), correctly-sized at District, which is exactly why one capture worked and the headline rung didn't. Diagnosed with live instrumentation (the tile-mode-rung suspect was ruled out with evidence but pinned with a named test anyway); fixed via pure zoom_compensated_size(); numeric pin of the pre-fix magnitude in the test suite.
  2. Byte-identical basins captures: root cause was the lead's drive script passing a button label where the API takes an overlay id — no product bug, but the chase banked a real regression pin (draw-counting spy proving toggles redraw the nature overlay).

Acceptance captures (.cache/screenshots/rivers-eyeball/)

R1: full river skeleton threading Lendel's continents at the orbital rest state, mouth rings on the coasts. R2: 7 watershed boundaries at the ruling's barely-there alphas. D: trunk-dim + mouth carve-out at a District window. Q: rivers correctly absent at Quarter.

Test discipline

Server: drainage 16/16 (+10 incl. the non-synthetic GJ1c trunk guard), layer1/layer_proxy/body_world_state green, cargo build --tests clean, revert-verified binning monotonicity. Client: 58+ new tests across four files (pure mapping/policy, lifecycle, real-driver draw smoke — with the smoke suite's vacuous-pass mode under X11 BadMatch honestly documented and the pure-function suite installed as the real gate), 11 existing atlas suites regression-free, three independent revert-verifications. Known accepted lint: atlas_window_viewer.gd at 1016/1000 advisory lines on trivial wiring — rides T-1158's decomposition rather than degrading its postmortem comments.

Tickets: T-1156 (wave 1). Filed from design pass: T-1168, T-1169, T-1170.

## Summary T-1156 wave 1: the nature layer's hydrology arrives on the Atlas zoom ladder — rivers, drainage basins, and attractors drawn over the derived terrain, per-rung filtered. Wave 2 (everything civilizational) remains explicitly deferred per the nature-first order. ### The design pass (before any code) Tyre's carrier ruling reframed the ticket: the river skeleton **already rides the wire** — `AtlasLayerResponse.layer1` has serialized `RiverNetwork`/`drainage_basins`/`attractors` whole-body since #952, and the retired draw code survives intact. So wave 1 is a *re-hosting* plus one additive field, touching **neither** D-226 ceiling (windowed family stays at one field; whole-body family untouched — `layer1` predates that framing). Per-rung refinement is **client-side class filtering** — deliberately unlike T-1162's server-side cutoff, because a fixed finite graph has nothing to Nyquist-truncate. The ruling establishes the wave-2 pattern for free: *discrete map features ride the whole-body overlay family filtered client-side; only continuous per-metre fields ride the windowed per-cell arrays.* Captured as a D-226 note on this branch. Design-pass honesty, ticketed not buried: **T-1168** (riparian vegetation response — `near_perennial_water` is a hardcoded-false stub, so rivers draw over barren-painted banks for now, deliberately not suppressed per Araminta+Tyre), **T-1169** (feature-name labels — three dormant pieces, not "cheap"), **T-1170** (the skeleton is 512×256 heightmap-resolution, ~76 km/cell — river *courses* below Region scale need invention, the linear analog of the coast crinkle). ### Server (Dudley) — `river_class` on `RiverNetwork` `river_class: Vec<u8>` (0=stream 1=tributary 2=trunk) parallel to `river_cells`, `#[serde(default)]`, computed at drainage-analysis time by log-scaled binning between `RIVER_THRESHOLD` and the **river-restricted** max accumulation. The restriction matters: grid-wide max includes below-sea-level ocean cells where accumulation peaks past the coastline — anchoring there starves the trunk class exactly on wet large-ocean bodies (caught in lead review of round 1; GJ1c went from zero trunk cells to a healthy 72/18/3 distribution). By construction every body with rivers has trunk cells, pinned by a non-synthetic regression on the committed GJ1c heightmap. `cascade_layer1.json` regenerated: every pre-existing field byte-identical (python-verified), only `river_class` added. ### Client (Stig) — the nature overlay on the ladder New `AtlasWindowNatureOverlay` node: self-connects to the shared `atlas_layers_received` broadcast (the established one-signal-N-consumers shape), consumes the whole-body skeleton, draws over the terrain composite. Pure tested coordinate chain `layer1_pixel → world_m → district → canvas-local` with the vertical convention **verified against the server source** (row 0 = north pole) and revert-verified pole tests. **Araminta's presentation ruling** (supersedes the provisional add-as-you-descend mapping, on the resolution evidence): Region/orbital = full skeleton (radii 0.9/1.4/2.2 px by class, confluence 3.5, mouth double-ring), District = trunk-only at 60%/1.6 px with mouths kept at full landmark styling, **Quarter = rivers off** until T-1170 — isolated 76 km dots there are placement coincidence, not information. Retired palette reused verbatim. Toggles RVR on / BAS off / ATR off. ### Live rounds (the eyeball check earning its keep again) 1. **Zoom-scaled marker sizes**: all marker radii were raw constants drawn inside the zoom-scaled canvas — sub-pixel-invisible at orbital fit zoom (~0.014 effective px), correctly-sized at District, which is exactly why one capture worked and the headline rung didn't. Diagnosed with live instrumentation (the tile-mode-rung suspect was ruled out with evidence but pinned with a named test anyway); fixed via pure `zoom_compensated_size()`; numeric pin of the pre-fix magnitude in the test suite. 2. **Byte-identical basins captures**: root cause was the *lead's drive script* passing a button label where the API takes an overlay id — no product bug, but the chase banked a real regression pin (draw-counting spy proving toggles redraw the nature overlay). ### Acceptance captures (`.cache/screenshots/rivers-eyeball/`) R1: full river skeleton threading Lendel's continents at the orbital rest state, mouth rings on the coasts. R2: 7 watershed boundaries at the ruling's barely-there alphas. D: trunk-dim + mouth carve-out at a District window. Q: rivers correctly absent at Quarter. ### Test discipline Server: drainage 16/16 (+10 incl. the non-synthetic GJ1c trunk guard), layer1/layer_proxy/body_world_state green, `cargo build --tests` clean, revert-verified binning monotonicity. Client: 58+ new tests across four files (pure mapping/policy, lifecycle, real-driver draw smoke — with the smoke suite's vacuous-pass mode under X11 BadMatch honestly documented and the pure-function suite installed as the real gate), 11 existing atlas suites regression-free, three independent revert-verifications. Known accepted lint: `atlas_window_viewer.gd` at 1016/1000 advisory lines on trivial wiring — rides T-1158's decomposition rather than degrading its postmortem comments. Tickets: T-1156 (wave 1). Filed from design pass: T-1168, T-1169, T-1170.
jpmschweitzer added 5 commits 2026-07-23 09:03:03 +02:00
docs(meta): D-226 wave-1 nature-overlay carrier note (T-1156 — Tyre's ruling) 365b044e79 
Rivers/basins/attractors ride the existing whole-body layer1 field —
neither AtlasLayerResponse ceiling touched, no tagged-envelope
migration. Per-rung refinement is client-side class filtering on the
additive river_class (distinct from T-1162's server-side Nyquist
cutoff: a fixed finite graph has nothing to truncate). General rule:
discrete map features ride the whole-body overlay family filtered
client-side; continuous per-metre fields ride the windowed per-cell
arrays — wave 2's roads/settlements inherit this unchanged. Riparian
vegetation response is the named T-1168 forward contract, deferred.

Tickets: T-1156

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
feat(simulation): T-1156 — quantized river_class on RiverNetwork (trunk/tributary/stream) ff98a10d42 
The one additive server change of wave 1 per Tyre's carrier ruling:
river_class: Vec<u8> parallel to river_cells, serde-default-safe, no
new wire field. Log-scaled binning between RIVER_THRESHOLD and the
RIVER-RESTRICTED max accumulation (max over cells passing the is_river
elevation filter — NOT the grid-wide max, which peaks past the
coastline on wet large-ocean bodies and would starve the trunk class
exactly where Araminta's District-shows-trunk-only table needs it;
caught in lead review of round 1, fixed round 2). Log not linear
because accumulation grows combinatorially downstream. By construction
every body with rivers has trunk cells — pinned by a non-synthetic
regression on the committed GJ1c heightmap at the golden's exact
downsample. cascade_layer1.json regenerated: every pre-existing field
byte-identical (python-verified), only the new river_class arrays
added — GJ1c distribution stream 72 / tributary 18 / trunk 3.
drainage 16/16 (+10), layer1 3/3, layer_proxy 50/50, build --tests
clean; revert-verified (non-monotonic binning fails the monotonicity
test by name).

Tickets: T-1156

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
feat(ui): T-1156 — rivers/basins/attractors re-hosted onto the zoom ladder (nature overlay node) 0d22e50e66 
New AtlasWindowNatureOverlay (Node2D on the viewer canvas, above
terrain): self-connects to the shared atlas_layers_received broadcast
(the established one-signal-N-consumers shape) and consumes the whole-
body layer1 skeleton per Tyre's carrier ruling — no windowed wire
touched. Coordinate chain layer1_pixel_to_world_m -> world_m_to_
district -> canvas-local lives in atlas_window_geometry as pure tested
functions; vertical convention (row 0 = North pole, wy increases
south) verified against the server's own pixel_to_world_m, pinned by
pole tests and revert-verified (sign flip fails them by name).
Araminta's per-rung presentation table implemented exactly: Region
full skeleton (radii 0.9/1.4/2.2, confluence 3.5, mouth double-ring),
District trunk-only 1.6px at 60% with mouths at full landmark styling,
Quarter off until T-1170 course invention; retired palette reused
verbatim; RVR on / BAS off / ATR off defaults; missing river_class
falls back to trunk. 58 new tests (34 geometry, 22 lifecycle, 2 real-
driver draw smoke actually rendered); 11 existing atlas suites
regression-free. atlas_window_viewer.gd runs 16 lines past the
advisory 1000-line cap on trivial wiring — accepted, rides T-1158's
decomposition.

Tickets: T-1156

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
fix(ui): T-1156 live rounds — zoom-compensated marker sizes; toggle-redraw regression pin 60faf667a5 
Live round 2 (the real bug): every nature-overlay marker size was a raw
screen-space constant drawn inside _canvas, whose scale IS view_zoom —
at Lendel's orbital fit zoom (0.0063) a 2.2px trunk dot rendered at
~0.014px, invisible; the same code at District's 3.75 zoom produced the
correctly-visible mouth ring, which is why one capture worked and the
headline rung didn't. Fixed via AtlasWindowGeometry.zoom_compensated_
size() (pure, floor-guarded) wired through every radius/line-width;
basin FILL points are positions and correctly stay unscaled. Suspect
tile-mode-rung-detection was ruled out live (granularity_v2=Region
confirmed in tile mode) but pinned with a named regression test anyway.
+8 pure-function tests incl. a numeric pin of the pre-fix magnitude
(<0.02px at orbital zoom); revert-verified by name. Draw-smoke suite
documented as supplementary (the shared SubViewport background harness
can pass vacuously under X11 BadMatch — the pure suite is the gate).

Live round 3 (drive-script bug, no product change): the lead's scratch
drive passed the button LABEL to set_overlay_visible() and the unknown-
id guard silently no-op'd — but the chase banked a real pin:
test_set_overlay_visible_gen_basins_flips_gate_and_redraws_nature_
overlay (draw-counting spy per the cold-start precedent; is_queued_for_
redraw does not exist in this build). Revert-verified.

Basins verified live: 7 Lendel watershed boundaries render at the
ruling's alphas. Suites: viewer 76/76, geometry-nature 42/42, nature-
overlay 22/22, zoom-ladder 50/50, no regressions across the cluster.

Tickets: T-1156

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
style(simulation): cargo fmt — assert message wrap in drainage tests 5a382c2e56 
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
jpmschweitzer added 3 commits 2026-07-23 09:33:32 +02:00
fix(ui): District trunk dot 2.0px/80% — Araminta's PR #195 capture objection 0366e8286f 
At 1.6px/60% the trunk dot under the mouth ring read as 'mouth glyph
over plain terrain' — invisible without knowing where to look,
underselling the ruling's own 'a major river crosses near here'
intent. 2.0px/80% preserves the fade-down ladder vs Region's
2.2px/100% without reading as accidentally-erased. Doc comment
records the revision provenance.

Tickets: T-1156

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
fix(ui): PR #195 review round — attractor stroke widths zoom-compensated (Tyre I1/I2/I3) 8a747e332c 
I1: _draw_attractor_shape's three stroke-width args (Confluence arc,
Coastal/NaturalHarbor arc, Oasis spokes) were raw screen-space literals
— Godot multiplies stroke widths by canvas scale exactly like radii, so
at the Region orbital fit zoom the outlines rasterized at ~0.01px, the
identical sub-pixel class the dot/ring compensation fixed, missed on
glyph internals (and attractors are Region-only — precisely where it
bites). Widths now arrive pre-compensated via a px_w param, keeping the
primitive pure. Regression pin: a source-scan test asserting no
draw_arc/draw_line in the function carries a bare numeric width (the
draw-smoke suite documents its own vacuous-pass mode, so source-scan is
the environment-independent gate); revert-verified by name. I2: D-226
visibility-direction sentence — Araminta's fade-down inversion recorded
as pre-T-1170 with its single revisit point named. I3: class-header
call-site claim corrected (enter() funnels through _enter_at_rung).

Tickets: T-1156

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
test(ui): named default-visibility pins for nature overlays (PR #195 Hoshe finding) c1cae7c9b5 
RVR-on-by-default is the single most player-visible behavior wave 1
ships (rivers appear with no toggle hunt) and was only exercised by
live captures; gen_basins' default was pinned incidentally inside the
toggle-redraw spy test and gen_attractors not at all. One named test
now asserts all three fresh-construction defaults per Araminta's
ruling (RVR on, BAS off, ATR off). Revert-verified: flipping the
gen_rivers init line fails exactly this test by name (1 failure total
in the 78-test suite). Suite 78/78; gdlint clean.

Tickets: T-1156

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
jpmschweitzer commented 2026-07-23 09:33:37 +02:00
Author
Owner
Copy Link
Copy Source

Review: rivers-ladder → main (type: code + visual sign-off)

Branch tip reviewed: c1cae7c9b (both fix rounds included). Reviewers: Tyre (architecture — reviewing the implementation of his own wave-1 carrier ruling), Hoshe (QA), Araminta (presentation sign-off on her own per-rung ruling).

Tyre (Architecture): APPROVE (after one fix round)

Initial REQUEST_CHANGES with one code find and two captures:

# Finding Resolution
I1 _draw_attractor_shape's three stroke-width args were raw literals — Godot multiplies widths by canvas scale like radii, so glyph outlines rasterized at ~0.01 px at the Region orbital fit zoom (the identical sub-pixel class the live-round dot/ring fix closed, missed on glyph internals — and attractors are Region-only, precisely where it bites) 8a747e332 — widths arrive pre-compensated via a px_w param (primitive stays pure); source-scan regression pin asserting no bare numeric width in the function (the environment-independent gate, since the smoke suite documents its own vacuous-pass mode); revert-verified by name
I2 Araminta's fade-down visibility inversion superseded the provisional add-as-you-descend mapping but lived only in code comments D-226 visibility-direction sentence: fade-down, explicitly pre-T-1170, RIVER_CLASS_VISIBLE_BY_RUNG named as the single revisit point
I3 Class-header call-site claim (doc nit) Corrected

His six architecture answers, in brief: the river-restricted anchor is ruled the correct model (not merely a bugfix — grid-wide max admits a ceiling no river cell can reach; per-basin anchoring recorded as a T-1170 forward option); the client-side const-table visibility is "the purest possible expression" of his carrier mechanism; the zero-ceilings claim confirmed definitively (empty diff on layer_proxy.rs — river_class nests inside the pre-existing whole-body layer1); the overlay's request-per-body-switch lifecycle is correct against D-227 "valid forever"; the D-226 note is faithful with no drift; and the divide-by-zoom compensation is defensible for wave 1 with an unscaled-sibling-layer refactor recorded for wave 2 (T-1158-adjacent, appended to T-1156).

Hoshe (QA): APPROVE (after one fix round)

Initial REQUEST_CHANGES on one gap: RVR-on-by-default — the single most player-visible behavior the feature ships — had no named regression assertion (gen_basins was pinned only incidentally; gen_attractors not at all). Fixed in c1cae7c9b: one named test pinning all three fresh-construction defaults, revert-verified to fail alone by name; Hoshe repeated the revert-verification independently. Full client gate re-run at the final tip: 3770/3770.

His verification was execution-grade throughout: own Python structural diff of the cascade golden (pure-append confirmed independently — only river_class added, 72/18/3 recomputed from the JSON); a forced clean rebuild proving no missed struct-literal site (the field is non-Option, so the compiler is the proof); his own sign-flip revert-test on the pole convention (exactly the two named tests fail); breaking the toggle-redraw wiring to confirm the spy test bites; and tracing cascade_golden.rs to confirm the GJ1c regression shares the golden's exact source and downsample. One tool incident self-caught and cleaned (a gdformat --check that mutates on this version; hand-restored byte-identical to HEAD).

Araminta (presentation sign-off): APPROVED, all four captures

R1 "delivers the read... dotted courses trace visibly across the continent"; R2's basin subtlety confirmed as designed ("tint, not paint" — do not strengthen); D initially objected — the District trunk dot at 1.6 px/60% read as accidentally-erased — fixed to her exact ask (2.0 px/80%, 0366e8286) and re-confirmed on a representative mid-latitude capture; Q's rivers-off validated in situ ("no false-precision artifact").

Live rounds during implementation (pre-review, for the record)

Two eyeball rounds preceded review: zoom-scaled marker sizes (sub-pixel-invisible at orbital fit zoom — diagnosed with live instrumentation, the wrong suspect ruled out with evidence and pinned anyway) and a byte-identical basins capture whose root cause was the lead's drive script passing a button label where the API takes an overlay id (no product bug; the chase still banked the toggle-redraw spy pin). One pole-adjacent entry wobble observed and filed as T-1171 rather than chased in-batch.

Verdict: APPROVED

Tickets: T-1156 (wave 1). Filed from design pass and review: T-1168 (riparian forward contract), T-1169 (labels), T-1170 (course invention + per-basin anchoring option), T-1171 (pole-entry wobble).

## Review: rivers-ladder → main (type: code + visual sign-off) Branch tip reviewed: `c1cae7c9b` (both fix rounds included). Reviewers: Tyre (architecture — reviewing the implementation of his own wave-1 carrier ruling), Hoshe (QA), Araminta (presentation sign-off on her own per-rung ruling). ### Tyre (Architecture): APPROVE (after one fix round) Initial REQUEST_CHANGES with one code find and two captures: | # | Finding | Resolution | |---|---------|------------| | I1 | `_draw_attractor_shape`'s three stroke-width args were raw literals — Godot multiplies widths by canvas scale like radii, so glyph outlines rasterized at ~0.01 px at the Region orbital fit zoom (the identical sub-pixel class the live-round dot/ring fix closed, missed on glyph internals — and attractors are Region-only, precisely where it bites) | `8a747e332` — widths arrive pre-compensated via a `px_w` param (primitive stays pure); source-scan regression pin asserting no bare numeric width in the function (the environment-independent gate, since the smoke suite documents its own vacuous-pass mode); revert-verified by name | | I2 | Araminta's fade-down visibility inversion superseded the provisional add-as-you-descend mapping but lived only in code comments | D-226 visibility-direction sentence: fade-down, explicitly pre-T-1170, `RIVER_CLASS_VISIBLE_BY_RUNG` named as the single revisit point | | I3 | Class-header call-site claim (doc nit) | Corrected | His six architecture answers, in brief: the **river-restricted anchor** is ruled the correct model (not merely a bugfix — grid-wide max admits a ceiling no river cell can reach; per-basin anchoring recorded as a T-1170 forward option); the client-side const-table visibility is "the purest possible expression" of his carrier mechanism; the **zero-ceilings claim confirmed definitively** (empty diff on `layer_proxy.rs` — `river_class` nests inside the pre-existing whole-body `layer1`); the overlay's request-per-body-switch lifecycle is correct against D-227 "valid forever"; the D-226 note is faithful with no drift; and the divide-by-zoom compensation is defensible for wave 1 with an **unscaled-sibling-layer refactor recorded for wave 2** (T-1158-adjacent, appended to T-1156). ### Hoshe (QA): APPROVE (after one fix round) Initial REQUEST_CHANGES on one gap: RVR-on-by-default — the single most player-visible behavior the feature ships — had no named regression assertion (gen_basins was pinned only incidentally; gen_attractors not at all). Fixed in `c1cae7c9b`: one named test pinning all three fresh-construction defaults, revert-verified to fail alone by name; Hoshe repeated the revert-verification independently. Full client gate re-run at the final tip: 3770/3770. His verification was execution-grade throughout: own Python structural diff of the cascade golden (pure-append confirmed independently — only `river_class` added, 72/18/3 recomputed from the JSON); a forced clean rebuild proving no missed struct-literal site (the field is non-`Option`, so the compiler is the proof); his own sign-flip revert-test on the pole convention (exactly the two named tests fail); breaking the toggle-redraw wiring to confirm the spy test bites; and tracing `cascade_golden.rs` to confirm the GJ1c regression shares the golden's exact source and downsample. One tool incident self-caught and cleaned (a `gdformat --check` that mutates on this version; hand-restored byte-identical to HEAD). ### Araminta (presentation sign-off): APPROVED, all four captures R1 "delivers the read... dotted courses trace visibly across the continent"; R2's basin subtlety confirmed as designed ("tint, not paint" — do not strengthen); D initially objected — the District trunk dot at 1.6 px/60% read as accidentally-erased — fixed to her exact ask (2.0 px/80%, `0366e8286`) and re-confirmed on a representative mid-latitude capture; Q's rivers-off validated in situ ("no false-precision artifact"). ### Live rounds during implementation (pre-review, for the record) Two eyeball rounds preceded review: zoom-scaled marker sizes (sub-pixel-invisible at orbital fit zoom — diagnosed with live instrumentation, the wrong suspect ruled out with evidence and pinned anyway) and a byte-identical basins capture whose root cause was the lead's drive script passing a button label where the API takes an overlay id (no product bug; the chase still banked the toggle-redraw spy pin). One pole-adjacent entry wobble observed and filed as T-1171 rather than chased in-batch. ### Verdict: APPROVED Tickets: T-1156 (wave 1). Filed from design pass and review: T-1168 (riparian forward contract), T-1169 (labels), T-1170 (course invention + per-basin anchoring option), T-1171 (pole-entry wobble).
jpmschweitzer closed this pull request 2026-07-23 09:35:29 +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#195