beezm 3b5bc5e5d6 docs: water at rest across the Core/Client/Server READMEs (terrain-water task 13)

Core: BlockRegistry gains WATER, with the reason it is wire-safe (block IDs
are never serialized) AND the reason it is a registry identity only rather
than a voxel the terrain path writes. Corrects the now-false line "Nothing
at runtime consumes the water data yet" -- it does, as of this task.
ChunkData's new per-column water fields and the Constants tunables.

Client: how the water sheet is built and, more importantly, WHY it is a
second mesh instead of a block in the density field -- the trap here is
that writing water into a Marching-Cubes field fuses it into the terrain
instead of laying it on top, and that is not obvious until you have done
it. Notes translucency came free from the separate material.

Server: the per-column water rule, which section is the authority for what
(WBID presence, WSRF level, WBTB fallback), that an unknown body leaves the
column dry rather than guessing, and why depth for shading comes from
blueprint heights rather than the clamped rendered geometry.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>

2026-08-09 02:33:06 -04:00

4.6 KiB

Raw Blame History

/Core/Scripts

Pure C# data structures, shared enums, and static maths. Compiled into the shared assembly used by both /Server and /Client.

"Pure data, zero state." These scripts define what things are and how to calculate them. They never track live game events — who is online, which chunks are loaded, what time it is.

What's actually here

Voxel materials

BlockData.cs — struct describing one block type (ID, name, IsSolid, BaseColor).
BlockRegistry.cs — the byte-ID master list (AIR 0, BEDROCK, STONE, DIRT, SAND, the three grasses, SNOW, WASTELAND_DIRT, ASPHALT, WATER) with a safe GetBlock lookup. Block IDs are never serialized — the blueprint stores heights, biome ordinals and water data, never block IDs, and chunks are not persisted — so appending to this table is wire-safe. WATER is a registry identity only: it is not written into ChunkData.BlockIDs, because water is drawn as its own surface rather than as part of the terrain iso-surface (see Client/README.md).
BiomePalette.cs — decides which material sits where, given biome, depth, and whether the column is roadbed. Two paths on purpose: an integer classification that decides what is stored in each voxel, and a float-depth version used for rendering that returns the two materials either side of a boundary so colours fade rather than snap.

⚠ Note that BlockRegistry and the mesher's colour table hold different RGB values for some blocks. The mesher's table is what you actually see; BlockData.BaseColor is currently unused by rendering.

Shared identifiers

Enums.cs — Biome, TownTier, MapHalf, RoadTier.

⚠ Biome and TownTier are the .dat wire format (u8 in the v2 container, i32 in legacy v1). They are declared without explicit values, so each member's ordinal is the number written to disk. Append only, at the end — never reorder or insert. The v2 reader range-checks ordinals so drift fails loudly at parse time, but only the append-only rule keeps old files meaning the same thing. Legacy v1 has no gate and no validation at all. RoadTier is exempt: road tiers live in separate sections of the file, so it is never serialized.

World data

BLUEPRINT_FORMAT.md — the byte-accurate .dat container contract (v2 tagged sections + the v1 legacy summary). Read this before touching the writer or parser.
BlueprintFormat.cs — the single registry of v2 constants: magic, version gate, section tags (FourCC), validation bounds, re-encode sentinels.
BlueprintWriter.cs — writes a WorldBlueprint as a v2 file. Blueprint-typed on purpose: the map generator and the round-trip harness are both just callers.
MapDataParser.cs — decodes a .dat blueprint into a WorldBlueprint (heightmap, biome map, towns, four road tiers, and — v2 only — the embedded generation params, per-town highway-node flag, and the optional water-bodies data: WaterBodyIds, WaterBodies table, WaterSurfaceQ). Dispatches on the first byte: v2 tagged-section files get validation (version gate, size bounds, section-length and ordinal range checks); legacy v1 files still load, intact, with a deprecation warning. The water sections are consumed at runtime since task 13 — the server reads them per column and the client draws the result (see Server/README.md).
ChunkData.cs — one chunk's density and block-ID fields, +1 padded on every axis so the mesher can reach into the neighbouring chunk, plus the per-column data the renderer needs — including WaterSurfaceY (world Y of the water surface, or Constants.NO_WATER), WaterDepthM (true depth from blueprint heights) and HasAnyWater.
Constants.cs — chunk dimensions, ISO_LEVEL, VOXEL_SCALE, and the visual/road tunables (material blend band; per-tier road width, shoulder, grade smoothing and surface), plus the water-at-rest tunables: HEIGHT_SCALE (the one raw-height→metres mapping, shared by the terrain surface and the water sheet so they cannot drift apart), NO_WATER, and the depth-shading colour/alpha ramp.

Maths

MarchingCubes.cs — density field → ArrayMesh, with analytical normals and deterministic integer-keyed vertex welding.
MATH_MARCHING_CUBES.md — an explainer of the algorithm.

Rules

No Godot node inheritance. Pure classes, structs and statics.
No _Process / _Ready. Nothing here is attached to a live scene object.
Dependencies flow inward only. /Server and /Client reference /Core; /Core never references them.

4.6 KiB Raw Blame History