Project Status

Last Updated: 2026-09-28 (Construction G3: foundation Add / Subtract. An editable blueprint is reshaped in place, an Add onto a built foundation becomes an extension blueprint that merges into it once built, and the result is judged against every foundation rule, including the new foundation-to-foundation pathing clearance. See dev log 2026-09-28-foundation-add-subtract.md. Earlier: 2026-09-27 WOR-493/WOR-488 fixed: the navmesh is one constrained Delaunay triangulation of the arrangement, so no face drops (chunk-corner water pinches and 300-tree land faces no longer blank the landing), a route deferred for want of a mesh is requested again when the nav generation moves, and the "Waiting for the area to settle" hold gives way to the first wander. See dev log 2026-09-27-frozen-colonist-unplaceable-stations.md. Earlier: 2026-09-26 Organic Terrain Geometry epic complete: water is the terrain rings for nav, rendering, vision, and mud, and every land boundary is a warped field isoline. See dev log 2026-09-26-organic-terrain-geometry.md. Earlier: WOR-463: perf validated across zoom levels; the ground and water pass stays under 1.5 ms of GPU at 3072x1728 at every zoom (now reported as tileGpuMs), far-zoom land LOD, 0.25 m near distance-field texels, the waterline lattice's one-sided cells filled in spans, nav regions recenter without blocking the frame. Earlier: WOR-462: every land-on-land boundary drawn as the isoline of a warped per-surface field (SurfaceField + land.glsl, golden-tested against each other), render tiles shrunk to 524 KB a chunk with a 3-tile apron post-processed like the neighbors' tiles, placement reading the painted field and the water rings, tile.glsl neighbor bleed and tile adjacency deleted. Earlier: WOR-461: vision drinks from shore points sampled along the rings, mud banks by distance to the rings that continue across chunk borders, point-bar sand on tight inner bends, TerrainPolygonQuery over each chunk's rings. Earlier: WOR-460: water and shore painted per pixel from the distance field, tile water path deleted, shader look tunable live over /api/dev/tunable. Earlier: WOR-459: nav reads the chunks' terrain rings, tile marcher deleted; colonists stop at the smooth waterline and wade fordable creeks. Earlier: WOR-457 per-chunk waterline rings built on the chunk worker, contour module in libs/geometry. Earlier: Game UI to Prototype Polish complete: 8 PRs merged (#256-#261, #263, #264) — layout verification harness (/api/ui/tree + /api/ui/lint), LayoutContainer auto-layout engine (Fill/Hug/Fixed, distribution, gap/padding, auto text wrap), MSDF text fixes, ScenarioSelect + PartySelect scenes with party→spawn wiring, starfield/planet backdrops, EntityInfoView rebuild, region minimap panel, dossier Gear paperdoll, full pre-game screen fidelity; lint reports zero violations on every screen. See dev log 2026-07-03-game-ui-prototype-polish.md. Earlier: Colonist task arbitration complete: (tier,score) key, 7-tier ladder, job lifecycle, TaskListView inspector, 879 engine-tests green, combined four-scenario flow verified end-to-end. Earlier: Colonist gameplay stabilization complete and merged (#241): grass is now a first-class GPU-instanced groundcover role driven by a procedural Lua asset (24 variants, ~486k tufts @ 120fps/1.75ms), placed entities no longer randomly tilt (opt-in ), and the EntityRenderer god-class was decomposed into per-path renderers. Story 0 of Living Environment Rendering; M-A wind next. See dev log 2026-06-27-groundcover-render-path.md. Earlier: Worldgen M7 complete: input-validation envelope + fail-loud reason channel, golden full-pipeline worldHash gate that turns dual-platform CI into a cross-platform determinism check (+ worldgen-cli --expect-hash), and a benchmark-calibrated default resolution (gen-time sweep on the reference machine; kept n=1024 as a documented kDefaultGridSubdivision). Closes the last engineering task of the World Generation & Creator epic; only the deferred planet-DB streaming and landing-difficulty UX remain. See dev log 2026-06-26-worldgen-m7-hardening.md. Earlier: Physical carry & stacks epics complete and merged (#218/#219/#224/#228/#230/#232/#233): two-hand wood armfuls + strength-derived carry, then a universal physical "stack" model -- one material capped at its own stackSize everywhere -- plus construction sites that hold their required materials directly and dump to the ground on demolish. See dev log 2026-06-26-physical-stack-inventory.md and plan 2026-06-26-bulk-material-carry-felling.md. Earlier: 2D landscape from 3D hydrology epic complete and merged (#212) -- replaced the noise-flood ponds with sparse biome/precip-weighted ponds + desert oases, implemented Distribution::Spaced with grove/glade/thicket clustering for denser natural forests, and fixed riparian plants (reeds on the mud bank + bankside bushes); follows the merged 2D rivers #208 and tributaries/springs #210. See dev log 2026-06-22-hydrology-ponds-and-proper-forests.md. Earlier: Realistic chained build loop: mass-based carry weight + axe-gated chopping + wood-unit tree harvesting; the colonist chops to its carry cap, hauls, repeats, then builds. Plus the task-chaining UI/AI fixes found verifying it (in-progress-work priority calc, friendly labels, full-width info-panel task lines, "Next" chain step, build-progress percent + on-map fill, global task list Build umbrella + who's-working-it). See dev log 2026-06-20-carry-weight-axe-gated-harvest.md. Earlier on main: Asset Manager epic #205 (dev log 2026-06-20-asset-manager.md), Cryosphere #199, Navigation P4 belief filtering #189/#191/#194, Vision System #172-184, Salvage UI cutover #176-181)

Epic/Story/Task Template

Use this template for all work items:

## Epic Title
**Spec/Documentation:** /path/to/spec.md or /path/to/folder/
**Dependencies:** Epic Name (if applicable)
**Status:** ready | in progress | blocked | needs spec

**Tasks:**
- [ ] Story Title
  - [ ] Task Title
    - [ ] Sub-task Title

Notes:

Max 3 levels of nesting (Story → Task → Sub-task)
Only last 4 completed epics + in-progress + planned epics shown here
Completed epics get a development log entry in /docs/development-log/entries/
An epic is only complete when ALL tasks are [x]

Recently Completed Epics (Last 4)

✅ Organic Terrain Geometry (terrain polygons)

Spec/Documentation: /docs/technical/organic-terrain/terrain-polygons-architecture.md, dev log 2026-09-26-organic-terrain-geometry.md Dependencies: none Status: complete (Specboard WOR-455; merged #268, #269, #274, #275, #276, #277, #278, #281, #282, and the WOR-463 PR)

Tasks:

✅ Game UI to Prototype Polish

Spec/Documentation: docs/design/ui/fidelity-gaps.md, docs/design/ui/ui-improvements-phase-2.md, dev log 2026-07-03-game-ui-prototype-polish.md Dependencies: Salvage UI Cutover Status: complete (merged #256, #257, #258, #259, #260, #261, #263, #264)

Tasks:

Result: The game UI matches the prototype's structure and polish on every screen, verified by machine-checked layout linting. Follow-ups recorded in the dev log: ASCII-only font atlas, font-family threading for wrapped text, window-size flag for multi-resolution lint, Hug-freeze engine wart. ✅

✅ Colonist task arbitration

Spec/Documentation: docs/technical/colonist-task-arbitration.md Dependencies: Colonist gameplay stabilization (complete), Goal-Driven Task Generation (in progress) Status: complete (3 commits on this branch)

Goal: Replace the drifted additive-float priority with a lexicographic (tier, score) key; add a job-lifecycle model so a colonist commits to an active work order across distractions.

Tasks:

Implement the (tier, score) lexicographic arbitrator in AIDecisionSystem
7-tier ladder classification for all action types (tiers in assets/config/work/priority-tuning.xml <TaskTiers>, fail-loud startup validation)
Within-tier score: distanceFactor (300*max(0,1-d/60), dominant), skillBonus, taskAgeBonus, hysteresis margin (50)
Job lifecycle: harvest locate-by-entity-id (kills phantom loop); StorageGoalSystem retires stale stocking-harvests; floor constants deleted
TaskListView inspector: tier/score/breakdown/selectionSummary, live update, T hotkey
Acceptance: combined four-scenario flow end-to-end, zero phantom harvest-loops, zero queued-job stalls, 879 engine-tests green
Docs updated (this file, ai-behavior.md, priority-config.md, colonist-task-arbitration.md, dev log entry)

Result: Lexicographic arbitration restored; phantom harvest-loop and queued-job stall fixed. Follow-ups: packaged-furniture lifecycle + findValidPositionNear; stocking re-arm / discovery-gating. ✅

✅ Colonist gameplay stabilization (craft → harvest → construction loop)

Spec/Documentation: dev log 2026-06-29-colonist-gameplay-stabilization.md; arbitration spec docs/technical/colonist-task-arbitration.md; testing suite docs/testing/ Dependencies: Navmesh/craft reliability (#240), Building & Construction Status: complete (merged PR #241)

Goal: Drive four end-to-end gameplay scenarios to reliable completion: craft axe, craft + stock storage box, fill a box to ~200 wood by real chop + haul, build a wood foundation with wall clearing + provisioning + construction.

Tasks:

Result: All four gameplay scenarios complete end-to-end. 872 engine tests green. Open follow-ups: crafted-box placement/packaged-furniture lifecycle, items-in-river/findValidPositionNear, mirrored colonist sprites, dev-tools spawn UX, arbitration implementation. ✅

✅ Physical carry & stacks (two-hand armfuls, universal stacks, construction-direct materials)

Spec/Documentation: dev log 2026-06-26-physical-stack-inventory.md, plan 2026-06-26-bulk-material-carry-felling.md; specs /docs/technical/bulk-materials-carry-and-felling.md, /docs/technical/physical-stack-inventory.md Dependencies: Realistic chained build loop (carry weight + axe-gated harvest) Status: complete (merged #218, #219, #224, #228, #230, #232, #233)

Goal: Make wood a two-hand bulky item carried as a strength-limited armful (felling is one destructive action that drops a loose pile), then make a "stack" a universal physical unit -- one material, quantity in [1, stackSize] -- everywhere from hands to boxes to build sites, with carry bound by weight or count, whichever comes first.

Tasks:

Bulk carry & felling (epic de719237, #218 + follow-ups #219): two-hand wood armful, Strength -> carry capacity, one-shot destructive felling, loose ground stacks, hand-aware haul/craft/deposit, HUD carry surfaces
Physical stacks (epic 0db82dab, #224/#228/#230/#232/#233): slot-based multi-stack Inventory, per-material stackSize (Wood 40, Berry 200, Stick 100), ground-pile split + scatter, construction sites hold materials directly (no Inventory; demolish drops salvage at tear-down completion), storage hauls sized by item capacity, armful clamped to stackSize

Result: A stack is one material in [1, stackSize] everywhere -- hands, pack, shelf, box, build site, ground; carry binds on weight or count, whichever comes first; build sites hold their manifest directly and dump to the ground on demolish. Deferred: per-stack container UI (88dc9e90), DrawingSystem rename (7d785fc6). ✅

✅ 2D Landscape from 3D Hydrology (rivers, ponds, forests)

Spec/Documentation: dev logs 2026-06-21-2d-rivers-and-water-landing.md, 2026-06-21-river-tributaries-springs.md, 2026-06-22-hydrology-ponds-and-proper-forests.md Dependencies: Water Availability (3D drainage), World Generation & Creator Status: complete (merged #208, #210, #212)

Goal: Render the 3D planet's hydrology into the 2D landing world at the right scale and make the surrounding landscape read as real -- followable rivers, sparse hydrology-driven ponds with desert oases, and dense naturally-clustered forests with working waterside plants.

Tasks:

2D rivers from 3D drainage (RiverNetwork2D: meandered flow-sized polylines, seam-safe) + water-biased landing (#208)
River tributaries, springs, and trickle headwaters (#210)
Hydrology ponds: remove noise-flood ponds; sparse biome/precip-weighted PondNetwork2D; desert oasis via distance-to-water (#212)
Proper forests: implement Distribution::Spaced; grove/glade/thicket density field; denser forest trees (#212)
Riparian plants: reeds on the mud bank (near="Mud", mud now forms on sand too); bankside berry/woody bushes in forest biomes (#212)

Result: The 2D world now grows rivers, sparse ponds, desert oases, clustered forests, and waterside plants from the 3D hydrology/biome data. Self-review on #212 fixed 4/5 findings; the 5th (cross-chunk Spaced spacing) is filed as Worldsim bug e0c87da0. Follow-ups: pond inlet/outlet streams, water-loving riparian trees. ✅

✅ Dev/Test Tools (HTTP verbs + developer-client tab)

Spec/Documentation: /docs/development-log/entries/2026-06-16-dev-tools-api-and-tab.md, /docs/development-log/plans/2026-06-16-dev-tools-api-and-tab.md Dependencies: None Status: complete (merged PR #157)

Completed Tasks:

Step 1: spawn, colonist, give (replaces givewood; site/loose/colonist/storage)
Step 2: need, time (speed/set/skip), teleport, select, kill, complete
Step 3: synchronous /api/state?what=summary|colonists|construction|time (screenshot-style request/response handshake)
Phase 2: Dev Tools tab (DevToolsService + DevToolsPanel) in the static developer-client
Refactor: dev surface extracted from GameScene into DevCommandHandler (scenes/game/dev/)
parseDevVerb tests refreshed (7/7); two rounds of Copilot review cleared (single-flight /api/state, CV wait, jsonEscape)

Result: Dev-only /api/dev/<verb> controls plus a synchronous /api/state readback, driven from a Dev Tools tab in the static developer-client, so sim state can be set up and asserted on without screenshots. In-browser tab click-test was skipped by choice (build clean, every endpoint curl-verified); demolish, inventory-in-readback, and query/count deferred as optional. ✅

✅ UI Design System (Salvage) — extraction

Spec/Documentation: /docs/design/ui/ Dependencies: None Status: complete (docs extracted from prototype; C++ implementation is a separate future epic)

Goal: Extract the approved "Salvage" UI prototype (docs/ui-prototype/) into durable design docs so the C++ build follows the docs, not the throwaway CSS. The prototype is kept as the live look-and-feel sandbox.

Completed Tasks:

Design language doc (principles, palette, typography, texture, motion)
Tokens: deterministic tokens.json (+ rgba floats, palette.svg) and readable tokens.md, generated from tokens.css
Component catalog: all 14 primitives (variants, props, states, anatomy, behavior/formulas)
Icons: 51-glyph set extracted to icons.json + icons-preview.svg + icons.md
Screen specs: 7 screens (pre-game flow + in-game HUD with every sub-state)
Mocks: rendered shots of every screen and in-game state (Playwright capture script) + anatomy wireframes for the dense screens
Re-runnable extraction scripts in the prototype; wired into /docs/design/INDEX.md

Result: /docs/design/ui/ holds the canonical UI spec. The data that only lived in code (icon paths, token values, component formulas, font identities) is now in text. Next: C++ implementation (generate constexpr Theme.h from tokens.json, build the primitive library, stroke-tessellated icons, multi-font). ✅

✅ Water Availability (worldgen + landing UX)

Spec/Documentation: /docs/development-log/entries/2026-06-15-worldgen-water-and-plate-realism.md, .claude/plans/water-hydrology.md Dependencies: Climate/Biome/Shelf Retune Status: complete (PRs #144, #146 merged)

Goal: Fix the coarse drainage so flow reaches the sea or ponds into basins, set the dead kFlagRiver/kFlagLake flags, and tell the player whether a landing site has fresh water (details pane + globe Hydrology mode). Worldgen stays coarse (amount + direction); actual river/lake geometry is deferred to chunk load.

Completed Tasks:

W-0: water stats baseline (WorldStats + worldgen-cli)
W-1: priority-flood depression routing + coarse kFlagRiver/kFlagLake + basin spill levels (resolution-invariant river quantile)
W-2: globe Hydrology color mode
W-3: landing-site water signal + details pane (river/lake on tile-or-neighbor, coast, rain-fed) + freshwater-biased default site
W-4: docs + PR; two Copilot rounds cleared; sub-1m lake spill-depth truncation fix (#146)

Result: Every land tile routes to the sea or a marked endorheic basin; lakes form in real basins; rivers ~4-5% of land, stable across resolution; landing pane reports water/terrain/climate/difficulty. Deterministic across thread counts. Follow-ups: valley erosion, 2D chunk-time river/lake rendering. ✅

✅ Plate Boundary Realism (straight-mountains fix)

Spec/Documentation: /docs/development-log/entries/2026-06-15-worldgen-water-and-plate-realism.md Dependencies: Tectonic History Simulation Status: complete (PR #145 merged)

Goal: Kill the dead-straight mountain walls that appeared across many generated worlds.

Completed Tasks:

Multi-agent diagnosis: root cause is the great-circle rift cut (tryRift plane-split), amplified by a flat plate-size distribution and a uniform belt-lift pedestal
Curved rift cut: two-front noisy Dijkstra split replacing the great-circle plane (oversized + suture paths)
Power-law plate-size hierarchy at seeding (few large, many small)
Along-belt lift modulation so ranges get passes/saddles instead of a uniform wall
Ultra 2048 generator option; golden hash re-pin; six-seed biome acceptance test; Copilot cleared

Result: Rift seams meander like real margins, no straight great-circle cuts on any seed (including low plate counts); the plate inventory has an Earth-like size hierarchy; mountain crests break into passes. Deterministic across thread counts. ✅

✅ Climate, Biome & Shelf Realism Retune (worldgen)

Spec/Documentation: /docs/development-log/entries/2026-06-14-climate-biome-shelf-retune.md Dependencies: Tectonic History Simulation Status: complete (PR #139 merged)

Goal: Retune the climate/biome stages (written against the old smooth terrain) for the new linear belts + bimodal hypsometry.

Completed Tasks:

C-1..C-5: meridional wind + distance-to-ocean moisture-advection sweep (rain shadows that scale to wide belts) + continentality temperature; piecewise continental shelf; biome rebalance + montane decouple + biome-fraction acceptance stats; dead-flag hygiene (kFlagCoast/kFlagGlacier); docs + dev log
Self-review (2 agents) + fixes; two Copilot rounds cleared; CI timeout fix (trimmed heavy acceptance tests)

Result: ArcticTundra ~47%→8-16%, HotDesert restored to ~8-10%, forests ~35-40%; visible rain shadows behind belts; shelves submerged ~6%; biomes bend around terrain instead of rigid latitude stripes. Deterministic across thread counts, n=1024 ~16.6s. Follow-up: water-availability hydrology, valley erosion. ✅

✅ Tectonic History Simulation (worldgen realism overhaul)

Spec/Documentation: .claude/plans/tectonic-history.md, /docs/development-log/entries/2026-06-13-tectonic-history-simulation.md Dependencies: World Generation & Creator (M1–M6) Status: complete (PR #136 merged)

Goal: Replace single-pass Voronoi plates + Gaussian-kernel terrain with a coarse time-stepped tectonic history (Euler-pole motion, subduction, ridges with crust age, collisions/sutures, rifting, slab pull, hotspots) upsampled to full resolution; elevation from Airy isostasy + seafloor depth-age law.

Completed Tasks:

M-T0–M-T6: headless worldgen-cli + WorldStats; PlateSim coarse sim; Wilson-cycle events (incl. M-T2.5 arc crust, M-T2.6 slab pull, M-T2.7 coherence); CrustStage + crustAge/orogenyAge + PlanetIO v3 (incl. M-T3.5/3.6 signed-distance coastlines); TerrainStage rewrite (incl. M-T4.5 linear belts + geodesic metric); deletion/hardening/edge cases; docs + dev log
Self-review (3 agents) + fixes; Copilot review cleared; flaky cancel test fixed

Result: Hexagonal continents and smooth dome mountains gone. Crisp organic coastlines, linear ridged mountain belts (geodesic aspect 5–7), bimodal hypsometry, Earth-like ocean ages, deterministic across thread counts, n=1024 in ~11s. Follow-ups: climate/biome retune, fluvial hydrology. ✅

✅ GameScene Subsystem Extraction

Spec/Documentation: /docs/technical/gamescene-refactoring.md Dependencies: None Status: complete

Goal: Extract placement and selection logic from GameScene into dedicated subsystems. Prevent god-class anti-pattern as gameplay systems grow.

Completed Tasks:

Result: GameScene reduced from ~900 lines to ~620 lines (placement: ~180 lines, selection: ~100 lines extracted). Subsystems now own their own state and logic. ✅

✅ Main Game UI: Core HUD

Spec/Documentation: /docs/design/main-game-ui-design.md (Sections 1, 2, 5, 9) Dependencies: Main Game UI: Primitives Foundation, UI Architecture: ViewModel Pattern Status: complete

Goal: Implement main gameplay HUD (top bar, colonist list, gameplay bar, zoom).

Completed Tasks:

Deferred:

Vertical scrollable colonist list (works fine at current scale, can add if needed)
A/Q/R hotkeys (conflict with WASD camera movement)

Result: Full HUD implementation with TimeSystem, TopBar with speed controls, GameplayBar with category dropdowns, enhanced colonist list with mood tints and double-click to follow, and floating zoom controls. ✅

✅ Simple Asset Support (SVG-Only)

Spec/Documentation: /docs/technical/asset-system/asset-definitions.md, /docs/technical/asset-system/folder-based-assets.md Dependencies: None (Folder-Based Asset Migration was done as part of this) Status: complete

Goal: Support hand-crafted SVG assets without Lua scripts via data-driven XML definitions.

Completed Tasks:

Optional Future Enhancements: Variation (colorRange, scaleRange), inheritance (parent attribute), LOD levels, shared SVG components (@components/).

Result: Assets like Reed, BerryBush, Berry, Stick, SmallStone, CraftingSpot all use this system. New simple assets can be added by creating a folder with XML + SVG. ✅

✅ Ground Texture System

Spec/Documentation: /docs/technical/ground-textures.md, /docs/development-log/plans/2025-12-22-ground-texture-system.md Dependencies: None Status: complete

Goal: Replace flat-color tiles with textured terrain using hybrid Tier 1 (rasterized texture) + Tier 3 (vector grass) approach, with hard/soft edge blending.

Completed Tasks:

Optional Future Work: Add missing patterns (Sand, Rock, Snow, Mud), enrich grass patterns, tune grass entity density.

Result: Tiles render with textured patterns, soft blending between same-family surfaces, hard edges at family boundaries (shorelines, cliffs). ✅

✅ UI Event System

Spec/Documentation: /docs/technical/ui-framework/event-system.md Dependencies: None Status: complete

Goal: Replace manual hit-testing boilerplate (~69 lines of QUICKFIX code) with event consumption. Components call event.consume() to stop propagation. MouseMove events handle hover state.

Completed Tasks:

Deferred: Debug Integration (hit-test endpoint, SSE events) - can be added when needed.

Result: Clean event propagation model. Adding new UI panels no longer requires updating multiple files. Components consume events to prevent click-through. ✅

✅ UI Architecture: Input System Cleanup

Spec/Documentation: /docs/technical/ui-framework/event-system.md Dependencies: UI Event System Status: complete

Goal: Complete the event-based input architecture. Remove handleInput(float dt) (wrong pattern per Game Programming Patterns) in favor of handleInput(InputEvent&).

Completed Tasks:

Architecture (Industry Best Practice):

while (running) {
    processInput();    // No dt - discrete events
    update(elapsed);   // dt here - advance simulation
    render();
}

Key insight: Input handling does NOT need delta time. It captures discrete events ("what happened?"). Delta time belongs in update() where simulation advances ("how much time passed?").

Result: Clean separation between discrete input events and continuous simulation updates. Scenes receive InputEvent& from Application and forward to UI components. Keyboard polling (WASD, ESC) happens in update(). ✅

✅ UI Architecture: FocusManager Simplification

Spec/Documentation: /docs/technical/ui-framework/focus-management.md Dependencies: UI Architecture: Input System Cleanup Status: complete

Goal: Reduce ~40 lines of FocusManager boilerplate per focusable component via CRTP base class.

Completed Tasks:

Create FocusableBase CRTP template (libs/ui/focus/FocusableBase.h)
- Auto-register in constructor
- Auto-unregister in destructor
- Handle move semantics correctly (unregister old, register new)
Migrate Button to use FocusableBase
Migrate TabBar to use FocusableBase
Migrate TextInput to use FocusableBase

Result: Components now inherit from FocusableBase<T> and use = default for move constructor/assignment. FocusManager registration is automatic. ~120 lines of boilerplate removed across 3 components. ✅

✅ UI Architecture: ViewModel Pattern

Spec/Documentation: /docs/technical/ui-framework/data-binding.md, /docs/development-log/plans/2025-12-24-viewmodel-pattern.md Dependencies: None Status: complete

Goal: Establish clear data ownership and efficient updates via ViewModel pattern.

Completed Tasks:

Architecture Notes:

Pull-based models: refresh(world) returns bool for change detection
Adapter layer: Centralizes ECS queries (ColonistAdapter, SelectionAdapter, CraftingAdapter)
UpdateType enum: None, Values, Structure, Show, Hide for tiered updates

Result: Clean separation between data (models), ECS queries (adapters), rendering (panels), and reusable UI elements (components). ✅

✅ UI Architecture: Theme System

Spec/Documentation: /docs/development-log/plans/2025-12-25-ui-theme-system.md Dependencies: None Status: complete

Goal: Centralize UI styling (colors, spacing, typography) via design tokens. Eliminate hardcoded color values across views.

Terminology:

Panel (design) = Visual style (dark bg, 1px border)
View (structural) = Composite UI element (EntityInfoView, TaskListView)
Component = Base class for all UI elements (unchanged)

Completed Tasks:

Deferred (Phase 2):

Edge mask support for docked panels (per-edge border control)
Per-corner radius when edges are removed

Result: All views now use centralized theme tokens from libs/ui/theme/. Hardcoded colors eliminated. Consistent visual identity across all panels. ✅

✅ UI Architecture: Layout System

Spec/Documentation: /docs/development-log/plans/warm-honking-hedgehog.md Dependencies: None Status: complete

Goal: Add automatic layout containers for component positioning via uniform layout API on all UI elements.

Design Decisions:

Uniform API on IComponent — All UI elements (components + shapes) participate in layout via getWidth()/getHeight()/setPosition() + margin
CSS-like margin — Margin adds to position (content draws inside) and adds to reported size
Single LayoutContainer — Direction attribute (Vertical/Horizontal) instead of separate VStack/HStack
No Spacer, no flex — Use margins for spacing between elements
Hybrid sizing model: Stacking axis is child-driven (query sizes), cross axis is parent-driven

Completed Tasks:

Result: LayoutContainer automatically positions children based on their reported sizes. Components use CSS-like margin for spacing. All UI elements participate in layout through uniform API. ColonistListView and TaskListView now use automatic layout instead of manual yOffset tracking. ✅

✅ Main Game UI: Complex Components

Spec/Documentation: /docs/design/main-game-ui-design.md (Section 17) Dependencies: ~~Main Game UI: Primitives Foundation~~ (complete) Status: complete

Goal: Build components for information-dense screens (resources, dropdowns, notifications).

Completed Tasks:

Result: All complex components implemented with theme-based styling, layout integration, and comprehensive test coverage (67 new tests total). Components use minimal constructors with sensible defaults. ✅

In Progress Epics

Build & Test Speed Improvements

Spec/Documentation: /docs/technical/build-performance.md Dependencies: none Status: in progress (Specboard epic; baseline measured 2026-07-03: CI gate 49 min, local MSBuild serial-per-project)

Goal: Ninja + compiler caching everywhere (ccache local, sccache CI), path-gated heavy worldgen tests, parallel/sharded ctest. Target: ~10-13 min CI gate on non-worldgen PRs; local clean builds in minutes, fresh-worktree builds mostly cache replay.

Tasks:

CI overhaul: sccache, Windows CI on Ninja, heavy bucket sharded + moved to path-gated tests-heavy.yml (PR #252, merged)
Local builds: Ninja Multi-Config + ccache presets, setup-msvc-env.ps1, foundation link hygiene (PR #253, merged)
PCH rollout after measuring cache interaction
Measurement week: fill spec Results table, revisit deferred items (lld-link, test selection)

Salvage UI Cutover (one unified UI library)

Spec/Documentation: /docs/design/ui/ Dependencies: UI Design System (Salvage) — extraction Status: core complete — every view/screen/widget styles off theme/Tokens.h; legacy Theme.h + PanelStyle.h deleted (PRs #164/#167/#171/#174/#175/#176/#177/#178/#180/#181, see dev log 2026-06-18-salvage-ui-cutover.md). Remaining: full dialog redesign (deferred).

Goal: One Salvage-styled UI library. Re-style the existing libs/ui widgets to the approved look via design tokens, fold the render-only design-system widgets into it, delete the duplicates, then rebuild the HUD, dialogs, and world creator to the prototype designs.

Tasks:

Fluvial Erosion (worldgen)

Spec/Documentation: .claude/plans/erosion.md, /docs/development-log/entries/2026-06-15-worldgen-fluvial-erosion.md Dependencies: Water Availability (drainage) Status: in progress (PR #149, draft)

Goal: Carve valleys into the continental terrain so rivers (rendered later from the climate drainage) land in real valleys. Detachment-limited stream-power incision (Braun & Willett implicit solver) on the drainage stack, between Terrain and Atmosphere; downstream climate/biomes/final drainage all see the carved terrain.

Tasks:

E-0: dissection metrics baseline (WorldStats + worldgen-cli)
E-1: ErosionStage core (shared DrainageRouting helper + implicit stream-power) + pipeline integration
E-2: downstream re-validation (carved terrain through climate/biomes/final drainage; determinism; budget)
E-3: strength at conservative subtle default (accepted)
E-4: dev log + status (this) on the draft PR
Mark PR #149 ready, clear CI + Copilot, merge

Result: Valleys carved deterministically (worldHash bit-identical across thread counts); mountains preserved (belt crests stay tall); world-tests green (174); gen ~22s at n=1024. Follow-ups: the bathymetry "comb" artifact (hex-BFS distance terracing on the shelf edge + crust-age depth field) is a separate deferred tuning task; then 2D chunk-time river/lake rendering.

World Generation & Creator

Spec/Documentation: /docs/design/features/world-generation/, .claude/plans/world-generation.md Dependencies: None Status: in progress

Goal: Real 8-phase world generation (plates → terrain → climate → biomes), WorldCreator scene with 3D planet view and landing site selection, wired into New Game → 2D gameplay sampling the generated world.

Tasks:

Building & Construction System

Spec/Documentation: /docs/design/game-systems/world/building-construction.md (design), /docs/technical/building-construction-architecture.md (architecture) Dependencies: Goal-Driven Task Generation (build/haul task sources); Navigation & Pathfinding before walls ship as gameplay — DONE (nav v1 landed 2026-06-16: walls/water/trees block movement, colonists route around them and through doors); Material economy — extend the early crafting implementation with choppable trees and construction material items (axe/chop work in flight; chains grow as systems need them) Status: in progress

Goal: Freeform polygon building: foundation blueprints drawn on the map, walls on foundations with snapping and edge-fill, doors/windows as parameterized vector assets, automatic room detection. Deliver-then-build loop with work-driven procedural construction visuals.

Tasks:

Goal-Driven Task Generation

Spec/Documentation: /docs/design/game-systems/colonists/task-registry.md, /docs/technical/task-generation-architecture.md Dependencies: ~~Task Ordering System~~ (complete) Status: in progress (core landed in PR #115; item reservations and Memory push integration remain)

Goal: Refactor task generation from discovery-driven (see item → create task) to goal-driven (goal exists → create task using Memory for fulfillment). This fixes the issue of thousands of invalid tasks appearing in the global task list.

Architecture Change:

Discovery → Updates Memory (what colonists know about)
Goals → Generate Tasks (what needs to be done)
Tasks exist at goal-level (~200), not item-level (~100,000)
Reservations are item-level (allows parallel work by multiple colonists)

Tasks:

✅ Storage and Hauling System

Spec/Documentation: /docs/design/features/storage-system.md, .claude/plans/storage-and-hauling.md Dependencies: ~~Main Game UI: Colonist Details Dialog~~ (complete) Status: complete

Goal: Allow crafting furniture (shelves, boxes), automatic hauling of loose items to storage.

Completed Tasks (All Phases):

✅ Main Game UI: Colonist Details Dialog

Spec/Documentation: /docs/design/main-game-ui-design.md (Section 8) Dependencies: ~~Main Game UI: Interaction Components~~ (complete), ~~Main Game UI: Primitives Foundation~~ (complete), ~~UI Architecture: ViewModel Pattern~~ (complete) Status: complete (merged PR #96)

Goal: Full colonist information display with 5 tabs, live updates while game runs.

Completed Tasks:

Dialog Structure (ColonistDetailsModel, ColonistDetailsDialog, GameUI integration)
Bio Tab (name, mood, current task)
Health Tab (8 needs as ProgressBars, mood summary)
Social Tab (placeholder)
Gear Tab (inventory items)
Memory Tab (TreeView with collapsible categories, scrollable)
Refactoring (tabs/ subfolder, TabStyles.h)

Completed MVP (Consolidated)

The following MVP epics have all been completed. Detailed task breakdowns are preserved here for reference:

✅ MVP Complete - All Core Systems Implemented

Status: complete (Dec 2025)

Summary: Full MVP implementation with autonomous colonist survival loop.

Completed MVP Epics:

Colonist Entity + ECS Foundation - Custom ECS with sparse set pattern, colonist spawning
Needs System - 4 needs (Hunger, Thirst, Energy, Bladder) with decay and thresholds
World Entities - Capability system, Berry Bush, Water Tiles, Bio Pile
Memory System - Vision-based entity discovery, no omniscient pathfinding
AI Decision System - Priority-based decision hierarchy (Tiers 3, 5, 7)
Movement & Pathfinding - Direct movement to targets with arrival detection
Actions System - Eat, Drink, Sleep, Toilet actions with side effects
Player Observation UI - Selection, need bars, task queue display

Key Technical Components:

ECS: EntityID with generations, ComponentPool, Registry, View, ISystem, World
Systems: VisionSystem, NeedsDecaySystem, AIDecisionSystem, MovementSystem, PhysicsSystem, ActionSystem, DynamicEntityRenderSystem
Components: Position, Rotation, Velocity, MovementTarget, Appearance, Colonist, NeedsComponent, Memory, Task, Action, DecisionTrace

Result: Colonist survives indefinitely, discovers entities through wandering, fulfills needs autonomously, and proactively gathers food when inventory is empty ✅

Deferred Epics

⏸️ Complex Flora CPU Optimization (deferred)

Spec/Documentation: /docs/technical/vector-graphics/animation-performance.md Dependencies: Vector Graphics Validation (completed) Status: deferred

Scope: Tier 2/3 CPU-side optimization (arena allocators, temporal coherence caching, SIMD Bezier flattening) for complex close-up world objects with true Bezier curve deformation (large trees), capped count, high zoom only. This is NOT the grass/groundcover path and NOT wind: dense groundcover renders via static GPU instancing and animates in the vertex shader (see Living Environment Rendering). Renamed from "Animated Vector Graphics Performance Optimization" because the old name collided with wind animation, which this epic does not own.

Reason deferred: MVP game systems take priority. GPU instancing already gives sufficient performance for current needs.

Remaining Work (deferred):

Phase 1: CPU Optimization Stack (arena allocators, temporal coherence, SIMD Bezier)
Phase 3: Tiered selector (route complex assets between the CPU and instanced paths by distance/complexity)

Planned Epics

UI Scale setting

Spec/Documentation: docs/design/ui/ui-scale-setting.md Dependencies: Settings screen Status: needs spec

Goal: One user-facing setting that scales the entire UI uniformly. The game renders the UI denser than the Salvage prototype (the logical viewport is ~2.5x the prototype's reference, so token-built elements come out ~19% of screen width where the mock is ~47%). A single global UI-scale factor applied at the logical-viewport level lets players size the UI to match the prototype's proportions, instead of hand-enlarging individual screens.

Tasks:

Calibration pass: measure current logical viewport vs the prototype's 1600x1000 reference; pick a default scale
Apply a single UI-scale factor at getLogicalViewport() derivation (layout, not bitmap post-scale); confirm DPI isn't already folded in
Verify pointer hit-testing stays aligned with the scaled draw space
Live re-layout on change (open dialogs/screens reflow, no reload)
Settings-screen control (slider/stepped) + persistence + reset; confirm MSDF text + tessellated icons stay crisp at the extremes

In-game dossier fidelity (remaining)

Spec/Documentation: docs/design/ui/screens/in-game.md ("Colonist details dialog"), prototype at docs/ui-prototype/src/screens/InGame/, mocks docs/design/ui/mocks/in-game-dossier*.png Dependencies: UI Scale setting (for absolute size); colonist systems below (for the data-backed tabs) Status: in progress (shell + Bio/Health/Memory/Tasks built; see PR #204)

Goal: Bring the in-game C++ ColonistDetailsDialog to full parity with the prototype. The dialog shell, persistent header, kicker/footer, 8-tab bar, and Bio/Health/Memory/Tasks layouts are built and match at the element level. Remaining:

Tasks:

Season System (season slider + dynamic seasonal snow)

Spec/Documentation: /docs/design/features/season-system.md (draft stub; background in docs/technical/cryosphere-ice-and-glaciers.md → "Season system") Dependencies: Cryosphere / ice (landed) Status: needs spec

Goal: Let the player view and select the time of year via a season slider, driven by real orbit mechanics (axial tilt + orbital position → per-tile insolation by date). Seasonal snow cover ("how much snow is on the ground right now") is DYNAMIC, computed at sim/view time from the date, NOT baked into worldgen — so changing the season must NOT require a full planet regeneration. Glaciers and the permanent-snow baseline stay in worldgen; seasonal snow is the thin, see-through layer on top of them.

Tasks:

Spec: orbit + axial-tilt model → per-tile insolation / temperature by date; how the date maps to the slider
Spec: dynamic seasonal snow (accumulate/melt from the date's temperature), separate from worldgen permanent snow + glaciers; where it lives (sim vs render), what is cached
Spec: how the globe and the 2D chunk views consume it without a planet regen
Season slider UI (view + select); recompute season without regenerating the planet
Seasonal snow render (the thin see-through layer the cryosphere doc defers)

✅ 2D Render Performance Overhaul (foundation, Phases 1-3 complete)

Spec/Documentation: .claude/plans/render-performance-overhaul.md, /docs/development-log/entries/2026-06-10-render-performance-overhaul.md Dependencies: None Status: core complete (Phases 1-3 landed 2026-06-10); Phases 4-5 parked as a small follow-up, not a blocker

Goal: Fix the measured collapse at zoom-out (4 FPS at 0.25x) and scroll hitches (112-443ms). Persistent GPU tile geometry, async entity bake, far-zoom impostor handoff. Target 120 FPS at every zoom at current flora density (headroom for massively increased density later); 4x density stress must hold 60+.

Foundation note: This epic established groundcover's two-representation contract: vector geometry up close, baked tile texture far away, with the impostor handoff between them at ~3px on-screen blade height. Living Environment Rendering builds directly on that contract.

Tasks:

Profiling tooling (camera/vsync control endpoints, perf-capture.ps1, draw call metrics fix)
Phase 1: Per-chunk tile-data textures (one quad per chunk; deleted per-frame drawTile path)
Phase 2: Async chunk generation + worker entity mesh bake + budgeted GPU uploads
Phase 3: Far-zoom impostor handoff (height-bucketed meshes, zoom cutoff + fade)
Windows frame pacing fix (timeBeginPeriod + sleep-then-spin; was capped ~60 FPS)
Phase 4: Remaining metrics correctness (GPU timer, Windows SystemResources, per-window max breakdown)
Phase 5: Small wins (AABB cache, LRU eviction, zoom-aware load radius, View smallest-pool, uniform-chunk fast path: skip the full 512x512 computeTile loop when all 4 sampled corners agree, e.g. deep-ocean chunks)
4x-density scroll hitch attribution (p99 64ms at 4x; within criteria but worth chasing before density increase lands)

Living Environment Rendering

Spec/Documentation: .claude/plans/living-environment-rendering.md, .claude/plans/we-have-a-number-fuzzy-pearl.md (grass capacity + groundcover plan) Dependencies: 2D Render Performance Overhaul (Phases 1-3, complete) Status: in progress — Story 0 (render path) shipped in PR #238 (merged 2026-06-27); reactive animation (M-A onward) continues, split in Specboard as the "Living Environment: reactive flora & water" epic.

Goal: Wind-blown grass at scale (potentially hundreds of thousands of blades), grass parting around colonists with footprint trails, animated reactive water. Vertex-shader animation + interaction displacement maps; no per-frame CPU tessellation. Prerequisite surfaced 2026-06-26: grass currently renders through the static baked path, which cannot hold that density (memory + upload). A dense-flora render path (static GPU instancing) must land before wind.

Tasks:

✅ Task Ordering System

Spec/Documentation: /docs/design/game-systems/colonists/task-registry.md, .claude/plans/task-ordering-system.md Dependencies: ~~Storage and Hauling System~~ (complete) Status: complete

Goal: Global task prioritization where all colony tasks are computed with numerical priority (int16), each colonist filters/modifies priorities based on skills/preferences, and in-progress/follow-up tasks receive priority bonuses.

Tasks:

Related Specs:

Task Registry — Core architecture
Priority Config — Tunable weights
Work Types Config — Work type definitions
Task Chains — Multi-step tasks
Task Generation Architecture — Technical deep-dive
Config Loading — When/how configs load (mod support)

UI Architecture: Animation System

Spec/Documentation: /docs/technical/ui-framework/animation-system.md (to be written) Dependencies: None Status: needs spec

Goal: Add tweening system for smooth UI transitions.

Background: The main-game-ui-design spec calls for smooth animations (modal fade-in, expand/collapse, toast slide-up). Currently no animation support.

Tasks:

✅ Main Game UI: Primitives Foundation

Spec/Documentation: /docs/design/main-game-ui-design.md (Section 17) Dependencies: ~~UI Architecture: Layout System~~ (complete) Status: complete

Goal: Complete scroll container and generalize progress bar for reuse across UI.

Completed Tasks:

Result: Generic progress bars and scrollable containers available in libs/ui/ for reuse across all UI. NeedBar simplified to thin wrapper. ✅

✅ Main Game UI: Interaction Components

Spec/Documentation: /docs/design/main-game-ui-design.md (Sections 8, 14, 17) Dependencies: ~~Main Game UI: Primitives Foundation~~ (complete) Status: complete

Goal: Build dialog, tooltip, and context menu systems.

Completed Tasks:

Result: All interaction components implemented with minimal APIs, theme-based styling, and comprehensive test coverage (54 new tests). Components use FocusableBase for keyboard handling. ✅

✅ Main Game UI: Information Systems

Spec/Documentation: /docs/design/main-game-ui-design.md (Sections 4, 6) Dependencies: Main Game UI: Complex Components, UI Architecture: ViewModel Pattern Status: complete

Goal: Add resources panel and notifications to the game UI.

Completed Tasks:

Deferred:

Resources Panel TreeView population (requires stockpile system)
Pin-to-always-show functionality

Result: Notification toasts appear in bottom-right with click-to-navigate. Resources panel shows collapsible "Storage" button with empty state. Z-order rendering fixed across all bars. ✅

Main Game UI: Minimap

Spec/Documentation: /docs/design/main-game-ui-design.md (Section 3) Dependencies: Main Game UI: Information Systems Status: planned

Goal: Add minimap for world overview and navigation.

Tasks:

Notes: Deferred from Information Systems epic due to GPU complexity. May require significant render-to-texture optimization work.

Main Game UI: Management Screens

Spec/Documentation: /docs/design/main-game-ui-design.md (Section 11) Dependencies: Main Game UI: Interaction Components (Modal), UI Architecture: ViewModel Pattern Status: planned

Goal: Full-screen management overlays.

Tasks:

Main Game UI: Polish & worldsim Features

Spec/Documentation: /docs/design/main-game-ui-design.md (Sections 12, 14, 16) Dependencies: Most other Main Game UI epics Status: planned

Goal: Final polish and worldsim-unique visualization features.

Tasks:

Vision System: Occlusion & Discovery

Spec/Documentation: /docs/technical/vision-architecture.md Dependencies: Geometry Foundations (libs/geometry); consumes Building & Construction's structure publication for occluders Status: complete — shipped #172/#173/#179/#183/#184, landed just before Navigation P4's belief filtering (its honest write path into memory). See dev log 2026-06-18-vision-system.md

Goal: Honest sight: GeometryIndex shared with navigation, visibility polygons with an outdoor fast path, discovery/witnessing/stale-memory reconciliation through visibility, windows pass sight while blocking movement, structures discoverable per segment. Must land with or before Navigation P4 (belief filtering is hollow without it). Fog of war explicitly excluded — separate later epic alongside the overlay system; the polygon data it needs comes free from this work.

Tasks:

GeometryIndex (segment store + transparency flags + version counter, shared with nav obstacle publication)
Visibility polygons (rotational sweep, outdoor fast path, per-observer caching)
Rewire discovery, witnessing, and stale-memory reconciliation through visibility
Structure-as-observable discovery + window transparency

Navigation & Pathfinding

Spec/Documentation: /docs/technical/pathfinding-architecture.md Dependencies: Geometry Foundations (shared integer-coordinate substrate, libs/geometry); consumes the construction obstacle/portal contract Status: in progress — navigation v1 shipped (P1, P2); belief filtering complete (P4, #189/#191/#194); regional layer complete (P3, 2026-06-23); P5/P6 remain. See dev logs 2026-06-16-navigation-v1.md, 2026-06-19-navigation-belief-filtering.md, and 2026-06-23-navigation-p3-regional-layer.md

Goal: Four-tier vector-native navigation: planet hex graph for cross-globe abstract parties, chunk connectivity components for reachability, dynamic CDT navmesh for exact local paths, collision circles + velocity-obstacle steering so agents take up space. Must land (through P4) before walls ship as player-facing gameplay; P6 unlocks raids.

Tasks:

P1: Agents become physical (radius component, dynamic spatial hash, circle collision, separation)
P2: Local navmesh, static world (exact CDT over the loaded area from terrain/water/flora/wall contours, triangle A* + corridor width + funnel, waypoint following). Shipped as one merged loaded-area mesh, off-thread rebuild on world change; integrated into MovementSystem + AIDecisionSystem; NavOverlay debug draw.
P3: Regional layer — DONE (2026-06-23). Width-aware reachability over the merged navmesh: connected components + Kruskal max-spanning reconstruction tree giving bottleneck(a,b) by LCA, in a truth forest (goal validity) and a terrain forest (belief-sound reject); the cheap reject (diameter > bottleneck, or different component) short-circuits the A* so failed searches vanish. Un-deferred Demyen corridor-width filtering (any agent size), grid point location, RRA* heuristic on the terrain graph (one resumable reverse search per goal serves every agent), A* instrumentation. Articulated narrow-cross-section routing model settled. Deferred: belief-dependent wall clearance, local dirty-section invalidation, NavOverlay/HTTP stats surface.
P4: Dynamic world + belief — DONE. Static slice (construction obstacles/portals, water, tree footprints, wall-collision safety net) plus belief filtering: one full-triangulation navmesh with per-face blocker tags, a per-colonist BeliefFilter consulted at query time (unseen walls treated as absent), beliefVersion-driven replans coalesced against vision, and nav-state readability in the colonist panel ("Going to" / "Re-routing" / "Can't find a way to"). Deferred: active search behaviors (door-discovery wall-follow, LKP search — NavState reserves their labels) and stale-demolished-wall pessimism.
P5: Crowds (velocity-obstacle avoidance + mitigations, occupancy costs, door slot queues, regression rig)
P6: Global tier + raids (hex-graph A*, abstract party records, attention bubbles, materialization handoffs, raider belief seeding + scouting)
Optional polish: brighter debug overlay, waypoint substepping for fast movers (nav-state line in colonist info panel DONE #194)

Simulation-Area Navigation LOD + Per-Asset Collision Geometry

Spec/Documentation: /docs/technical/vector-graphics/collision-shapes.md, /docs/technical/pathfinding-architecture.md, /docs/development-log/entries/2026-06-24-sim-area-nav-lod-and-collision-geometry.md Dependencies: Navigation & Pathfinding (P3 regional layer) Status: complete (A-G + F + D2 shipped: #217, #220, #221, #225, #226, #227)

Why: nav-v1 never built the navmesh in a forested world (every loaded tree carved into one ~88k-poly mesh; the off-thread build never finished, colonists beelined). Fixed architecturally with a viewport-anchored simulation-area LOD; the long-specced per-asset collision geometry landed alongside.

Tasks:

A/B/C: Simulation-area nav LOD — viewport-anchored navmesh build (forested world builds <1s), pan/zoom tracking (capped 64m, rebuild on in-area placement), off-area beeline seam (#217)
D1: Rect collision primitive — drop Circle; rect-or-polygon; XML <rect> + Lua asset:setCollisionRect (eager capture at load); trees emit trunk rects (#220)
E: Tier-3 static-rect collision — StaticRectCollisionSystem pushes agents out of trunk rects at the nav-pad clearance (#221)
G: Asset-manager collision overlay + larger preview (#225)
F: Spec → Implemented + dev-log + flora-skill collision docs
D2: SVG-metadata collision authoring + BigRock obstacle (simple assets author the collider in SVG <metadata>) (#227)
Follow-ups (filed): visual-size obstacle LOD (lift the 64m cap); unify procedural collider frame; coarse far-field nav layer (LOD1)

Flora Content Pack

Spec/Documentation: /docs/technical/asset-system/ Dependencies: ~~Simple Asset Support~~ (complete) Status: ready

Goal: Add visual variety with flowers, mushrooms, rocks, and bushes.

Variant Cache System

Spec/Documentation: /docs/technical/asset-system/README.md (World Seed section) Dependencies: Folder-Based Asset Migration Status: planned

Goal: Pre-generate and cache procedural asset variants for fast loading.

Observability System - Full Feature Set

Spec/Documentation: /docs/technical/observability/INDEX.md Dependencies: None Status: planned

GPU-Based SDF Rendering for UI Primitives

Spec/Documentation: /docs/technical/ui-framework/sdf-rendering.md Dependencies: None Status: planned

Mod Support

Spec/Documentation: /docs/technical/asset-system/mod-metadata.md, /docs/technical/asset-system/patching-system.md Dependencies: Folder-Based Asset Migration, Variant Cache System Status: planned (post-MVP)

Blockers & Issues

Known limitations — chained build loop (non-blocking)

From the realistic-harvest work (dev log 2026-06-20-carry-weight-axe-gated-harvest.md):

Multi-colonist carry accounting is coarse. ConstructionSystem::carriedAmount sums Wood across all colonists, while the per-trip bound is a single colonist's. Fine for the single-cutter case; revisit when balancing load across colonists.
Dev/tool-placed entities don't survive chunk unload. If a colonist roams far with no work and the origin chunks unload, dev-spawned trees and committed foundations are lost, leaving the colonist idle. World-gen flora regenerate on chunk reload, so this is mainly a dev-test artifact; a colonist with genuinely no reachable work also wanders far (cosmetic).
Builds are slow-ish. ~30s of a colonist standing in place for a 36 m² foundation at normal speed (work scales 12 units/m²). Now legible via the percent + on-map fill; lower workRatePerSquareMeter if a snappier feel is wanted (balance dial, not a bug).
"Next" chain step verified live only for the construction chain. Craft/storage chains are handled in code (destination classification) but not yet exercised live.

✅ RESOLVED: SVG Ellipse/Circle Tessellation Bug

Impact: Berry Bush and other assets using <ellipse> or <circle> SVG elements fail to render Resolution: Fixed in commit e246fc0 (Dec 7, 2025) by adding convex polygon detection + fan tessellation

Added isConvexPolygon() to detect circles/ellipses (inherently convex)
Convex polygons use O(n) fan tessellation, bypassing the problematic ear-clipping algorithm
Unit tests added in Tessellator.test.cpp to prevent regression
Berry Bush SVG now renders correctly with native <circle> elements

✅ RESOLVED: Entities Spawning on Water Tiles

Impact: Flora entities (grass, berry bushes, trees) spawn on water tiles instead of being restricted to land Resolution: Fixed by Flat Tile Storage Refactor + Retina Coordinate Fix

Flat tile storage gives single source of truth for surface type
Retina coordinate fix ensures correct world↔screen mapping
Colonists now correctly find and drink from actual water tiles
No visual grass-on-water issue observed

MVP Success Criteria

From /docs/design/mvp-scope.md:

Test Scenario:

One colonist spawns at map center
Several berry bushes scattered nearby
Water tiles (ponds) within walking distance (tile-based, not entity)
Open ground for sleeping/bathroom

Expected Behavior (leave running):

Colonist wanders initially
Discovers berry bush and pond through sight
When hungry → walks to known berry bush → eats
When thirsty → walks to known pond → drinks
Drinking increases bladder need
When bladder urgent → finds outdoor spot → creates Bio Pile
When tired → sleeps on ground
Repeat forever

Success:

Colonist survives indefinitely
Task queue shows sensible decisions
No player intervention required
Colonist discovers new entities through wandering

FilesExpand file tree

status.md

Latest commit

History

status.md

File metadata and controls

Project Status

Epic/Story/Task Template

Recently Completed Epics (Last 4)

✅ Organic Terrain Geometry (terrain polygons)

✅ Game UI to Prototype Polish

✅ Colonist task arbitration

✅ Colonist gameplay stabilization (craft → harvest → construction loop)

✅ Physical carry & stacks (two-hand armfuls, universal stacks, construction-direct materials)

✅ 2D Landscape from 3D Hydrology (rivers, ponds, forests)

✅ Dev/Test Tools (HTTP verbs + developer-client tab)

✅ UI Design System (Salvage) — extraction

✅ Water Availability (worldgen + landing UX)

✅ Plate Boundary Realism (straight-mountains fix)

✅ Climate, Biome & Shelf Realism Retune (worldgen)

✅ Tectonic History Simulation (worldgen realism overhaul)

✅ GameScene Subsystem Extraction

✅ Main Game UI: Core HUD

✅ Simple Asset Support (SVG-Only)

✅ Ground Texture System

✅ UI Event System

✅ UI Architecture: Input System Cleanup

✅ UI Architecture: FocusManager Simplification

✅ UI Architecture: ViewModel Pattern

✅ UI Architecture: Theme System

✅ UI Architecture: Layout System

✅ Main Game UI: Complex Components

In Progress Epics

Build & Test Speed Improvements

Salvage UI Cutover (one unified UI library)

Fluvial Erosion (worldgen)

World Generation & Creator

Building & Construction System

Goal-Driven Task Generation

✅ Storage and Hauling System

✅ Main Game UI: Colonist Details Dialog

Completed MVP (Consolidated)

✅ MVP Complete - All Core Systems Implemented

Deferred Epics

⏸️ Complex Flora CPU Optimization (deferred)

Planned Epics

UI Scale setting

In-game dossier fidelity (remaining)

Season System (season slider + dynamic seasonal snow)

✅ 2D Render Performance Overhaul (foundation, Phases 1-3 complete)

Living Environment Rendering

✅ Task Ordering System

UI Architecture: Animation System

✅ Main Game UI: Primitives Foundation

✅ Main Game UI: Interaction Components

✅ Main Game UI: Information Systems

Main Game UI: Minimap

Main Game UI: Management Screens

Main Game UI: Polish & worldsim Features

Vision System: Occlusion & Discovery

Navigation & Pathfinding

Simulation-Area Navigation LOD + Per-Asset Collision Geometry

Flora Content Pack

Variant Cache System

Observability System - Full Feature Set

GPU-Based SDF Rendering for UI Primitives

Mod Support

Blockers & Issues

Known limitations — chained build loop (non-blocking)

✅ RESOLVED: SVG Ellipse/Circle Tessellation Bug

✅ RESOLVED: Entities Spawning on Water Tiles

MVP Success Criteria