Commit graph

2 commits

Author SHA1 Message Date
6d04c8229b Phase 5 visual polish: real 3/4 walls, bigger cabin, subtle grass
User feedback after first visual inspection: 'walls looked really thick
and there was hardly any inside space'. Five targeted fixes:

1. Wall._draw rewrite — proper within-tile 3/4 perspective. Previously the
   wall was a 16w×32h bottom-anchored rect that rose UP into the cell
   above (which encroached on the cabin interior whenever the wall faced
   south). New layout: 16×16 fitting strictly inside its own tile, with
   a lit-top 5px band + shaded-front 11px band + mortar lines + outline.
   Reads as 'a wall standing up' without overlapping adjacent cells.
   Same shape for stone and wood materials (different palettes).

2. Demo cabin expanded from 5×4 to 8×6 — interior went from 6 cells
   (closet) to 24 cells (actual room).

3. Grass-tile border darken: 0.15 → 0.04. Killed the graph-paper effect
   that dominated at 3×+ zoom. Tile boundary still readable when looking
   carefully; doesn't dominate the visual.

4. Removed Phase 1 _paint_sample_walls() seed. That 8×8 stone ring lived
   only on the Wall TileMap layer (set to visible=false in the Phase 5
   rendering pivot) — so it was an invisible path obstacle. Cleaner to
   not seed it at all now that walls render at entity level.

5. CameraRig default target_zoom: 1.0 → 2.5. At 1× on a 1280-px viewport
   the world feels sparse and pawns become 6-pixel dots. 2.5× shows ~30
   tiles wide which is the 'comfortable inspection' level.

MCP-verified visually: the cabin now reads as a proper 8×6 stone room
with raised 3/4 walls and a real 24-cell interior. Subtle grass field.
The 'commit to 3/4 perspective' rendering decision is now actually
visible (was conceptual-only before).

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
2026-05-10 22:30:40 +01:00
f82807ff3d Phase 5 — Designation, BuildJob, Wall/Floor/Door, Crate; 3/4 perspective pivot
Three gdscript-refactor agents in parallel + Opus integration.

Architectural pivot (memory.md Decisions table updated):
- View: top-down grid for gameplay + 3/4-perspective rendering for vertical
  structures (Stardew/Going Medieval style). Walls/doors/crates are Y-sorted
  entity sprites, not TileMap cells.
- Wall TileMap layer (Layer 2) becomes data-only — used for room detection,
  roof BFS, save serialization. Visual rendering happens at entity level.
- Asset reality check baked into the decision: the entire asset library is
  RPG-style perspective art; pivoting the renderer is cheaper than authoring
  or commissioning top-down 47-tile autotile sets.

Designation paint (scenes/world/, Agent A — ~170 lines):
- class_name Designation extends Node, lives as DesignationCtl child of World
- TOOL_NONE / TOOL_BUILD_WALL / TOOL_BUILD_FLOOR
- _unhandled_input captures left-mouse press/drag/release
- Drag-paints ghost tiles on Layer 3 via paint_layer.set_cell
- Green/red modulate based on World.pathfinder.is_walkable + cell occupancy
- Emits EventBus.designation_added/cleared per cell
- Selection.designation_active guard prevents double-handling clicks

EventBus signals added:
- designation_added(cell: Vector2i, tool: StringName)
- designation_cleared(cell: Vector2i)

BuildJob + Wall/Floor/Door entities (scenes/ai/ + scenes/entities/, Agent B — ~530 lines):
- Toil.KIND_BUILD + Toil.build_at(target_path) factory
- JobRunner._tick_build: resolves NodePath target, calls on_build_tick() per
  sim tick, marks toil done when is_buildable() returns false
- ConstructionProvider (priority=6, highest): nearest is_buildable() site in
  World.build_queue → Job=[walk_to(site.tile), build_at(site.get_path())]
- Wall entity: BUILD_TICKS=100, 40% alpha ghost; _complete() calls
  pathfinder.set_cell_walkable(false) + World.mark_wall_tile + Audit.log
- Floor entity: BUILD_TICKS=30, ground-level (no y_sort), does NOT block
  pathfinding, calls mark_floor_tile on complete
- Door entity: BUILD_TICKS=80, bottom-anchored, walkable when built (no
  pathfinder block); registers with World.doors for Phase 7 open/close logic
- ALL wall/door scenes have y_sort_enabled=true on root; floors don't (always
  on ground plane)

Crate furniture (scenes/world/, Agent C — ~270 lines):
- class_name Crate extends StorageDestination (Phase 4's abstract base)
- CAPACITY=4 stacks; accepts() gates on _completed + _filter_accepts + room
- find_drop_position returns tile when room exists, (-1,-1) otherwise
- BUILD_TICKS=60; on_build_tick mirrors Wall's pattern
- _draw procedural brown crate body + slat lines + fill-level dots

World autoload extensions (Opus):
- build_queue: Array — Wall/Floor/Door/Crate ghost entities awaiting
  construction. ConstructionProvider iterates by .priority desc; Phase 6+
  prepends material-haul toils.
- doors: Array — completed doors for future open/close (Phase 7+)
- wall_layer / floor_layer / designation_layer refs exposed for entity code
- mark_wall_tile(tile, material) / mark_floor_tile(tile, material) —
  stamps data-only TileMap layer with material-encoded atlas coord
- stockpile_at_tile(tile) — finds StockpileZone OR Crate covering a tile;
  used by JobRunner._tick_deposit to route Crate deposits

JobRunner._tick_deposit refactor (Opus):
- After clearing the haul-dirty flag, looks up stockpile_at_tile(pawn.tile)
- If destination is a Crate (has_method('register_item')), calls
  dest.register_item(item) so the crate's _contents tracks the stack

World scene integration (Opus):
- y_sort_enabled=true on World root so all entity sprites sort correctly
- DesignationCtl, ConstructionProvider, Wall TileMap (data-only, visible=false)
- World._ready wires:
  * World.wall_layer / floor_layer / designation_layer
  * designation.bind(designation_layer, selection)
  * Register 5 work providers (construction=6 > chop=5 > mine=4 > haul=3 > rest=0)
  * EventBus.designation_added → _on_designation_added (spawns Wall/Floor entity)
- _seed_phase5_demo_buildings: pre-queues 14 wall designations forming a
  5×4 cabin outline at (45, 25) so pawns visibly construct walls without
  player-paint UI (deferred to Phase 17). Spawns 2 fully-built crates at
  (17-18, 60) for hauling routing.

Acceptance — MCP-verified end-to-end:
- 14 wall designations seeded at boot, 2 crates pre-built
- All 3 pawns picked construction (highest priority work) and walked to
  build sites (paths 37/32/27 from spawn). Walls built one by one.
- Wall layer post-construction has 42 cells: 28 (Phase 1 stone ring) +
  14 (Phase 5 cabin) — both rendering paths (placeholder TileMap from
  Phase 1, plus new entity sprites from Phase 5) coexist correctly.
- Pathfinder set_cell_walkable(false) fired on each wall completion.
- Pawns transitioned from construction to hauling once all walls done.
- Final visual: 5×4 stone-walled cabin with mortar lines, Y-sorted entity
  rendering, wood items scattered east of the cabin awaiting haul.

Phase 5 gotchas (logged):
- 'material' as a member var shadows CanvasItem.material (Node2D inherits
  it). Renamed to wall_material / floor_material via quick-edit agent.
  Save-format dict KEYS stay as 'material' for stability.
- Class-name registration cache lag bit again (Tree/Pawn pattern from
  earlier phases). Workflow stays: agent writes class_name file → MCP
  reload_project → godot --headless --editor --quit → headless validate.
- ConstructionProvider scans build_queue every tick including completed
  walls; is_buildable() filters them out but the queue keeps growing.
  Phase 16+ should add an unregister_build_site call from _complete or
  a periodic queue compaction.

Delegation report this phase:
- Agent A (Sonnet, gdscript-refactor): Designation paint mode + EventBus
  signals + Selection guard. ~180 lines.
- Agent B (Sonnet, gdscript-refactor): Toil.KIND_BUILD + JobRunner._tick_build
  + ConstructionProvider + Wall/Floor/Door entities + scenes. ~530 lines.
- Agent C (Sonnet, gdscript-refactor): Crate furniture extending
  StorageDestination. ~270 lines.
- quick-edit (Haiku): material → wall_material/floor_material rename. ~14
  occurrences across 2 files.
- Opus: World autoload extensions + JobRunner _tick_deposit refactor +
  World scene integration (DesignationCtl + ConstructionProvider + new
  scene preloads + _seed_phase5_demo_buildings) + MCP runtime verification
  + the material-shadow + class-cache-lag debugging.

Pivot decision worth flagging: the asset library audit revealed that no
pack we own ships top-down 47-tile autotile walls. After multiple
researcher-overpromise cycles, the pragmatic call was to pivot the
rendering model itself. Walls now render as bottom-anchored tall sprites
with Y-sort; the entire asset library becomes usable as-is. Phase 17
polish will swap procedural _draw() with AtlasTexture regions from
Pixel Crawler / FG_Houses / Ventilatore Castle_Building.

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
2026-05-10 22:11:40 +01:00