Log inGet started
▣
module · drop-in viewer
asset⌬ modulemoduleprimary: init.luau·part ofmodule shared.module·originates fromworld 07158574-5…

saver

Luau-side scene saver. Writes `scene.json` v6 — **authored intent only**. No captured runtime state.

byzero-proxy @ DESKTOP-DB3UJOJ·posted 4mo ago
What it does

scene.shared.saver

Luau-side scene saver. Writes scene.json v6 — authored intent only. No captured runtime state.

Public surface

local saver = require("@builtin::assetTypes.scene.shared.saver")

saver.save(name, opts?)
  -- name: logical scene name ("main") OR full VFS path
  -- opts.layer: layer name to read from (default layers.active.name or "main")
  -- Returns: the absolute VFS path written.

What is captured

FieldSourceNotes
format, versionconstantsAlways "scene" + 6.
entitiesentity.findAll() walkExcludes temporary entities + the spawner-managed Player + primary Camera.
entities[].transformlocal position / rotation / scaleNOT world transform — local only.
entities[].componentscomponent.list + component.getStrips _-prefixed runtime fields.
entities[].attributesattribute.list + attribute.getThe entity's key/value bag, written as a { [key] = value } map; omitted for an entity that holds none. Holds the authored keys — a _-prefixed key is the engine's own session bookkeeping.
lightinglights.snapshot(), filteredScene-level clear_color + sky only. Directional / ambient / point lights persist as Light entities in entities[].
playersceneProxy.player_configCarried forward from load-time, authored only.
camerasceneProxy.camera_configCarried forward from load-time, authored only.

What is NOT captured

  • Local player's position / rotation. Runtime drift; not authored intent.
  • Primary camera's transform. Runtime; behavior-driven.
  • Any entity flagged temporary = true. Debug helpers, runtime markers.
  • Internal component fields prefixed with _ (e.g. _velocity on CharacterController).

The spawner-managed entities (Player + primary Camera) are excluded because they're created on load by player_spawner / camera_spawner from the scene's declarative player + camera blocks. Saving them would double them on the next load.

Parts

The module is a folder of parts, each a module of its own that init.luau requires:

  • scene_saver.serialize: one live entity as the record a scene file carries.
  • scene_saver.recordDiff: the comparisons the record diffs are built from.

When to use

layers.active:save(name) (Phase 6) routes through this module. Callers that want a state snapshot (save-game systems, undo stacks) should use world.debugSaveToDisk(...) instead — that captures runtime state by design.

Interface

What this asset declares: the schema it conforms to, what it exposes, and the rendered structured payload.

conforms to

zero/source-extract/v2

module scene_saver Luau-side scene saver. Persists scene state via a delta-overlay model: Canonical (committed work) <scene folder>/scene.json Dirty (work-in-progress, deltas only) <scene folder>/scene_dirty/manifest.json scene-level config overrides (lighting, player, camera, format, version) <scene folder>/scene_dirty/entities/<id>.json one file per CHANGED entity: * full body — modified or newly spawned * tombstone — { tombstone = true } for despawn The dirty directory carries ONLY the deltas. Entities unchanged since the last canonical save have NO entry under entities/. Editing one cube in a 100k-entity scene writes exactly one ~1KB file per drain; the manifest is touched only when scene-level config changes. The entity-id list is NOT carried in the manifest — it's derived at load / promote time from `canonical.entities[]` plus a filesystem listing of `entities/`. Concurrent writes to per-entity files therefore never produce manifest-conflict orphans (the conflict surface is one entity at a time, not the whole-scene index). Tombstones survive in dirty until the next save (when `clearDirty` wipes the directory). An orphan tombstone — one targeting an entity that wasn't in canonical either (i.e. the entity was spawned and deleted inside the same edit session before save) — is a no-op at compose time, so the file shape is robust regardless of operation order. The composer (`composeMerged`) is the canonical merge point used by both this saver's promote/save paths AND `scene_loader.M.load`, so there's exactly one place that knows how to combine canonical with the overlay. Captures authored intent only: no spawner-managed entities (player identities, primary camera), no temporary entities, no runtime player position. FFI shapes (discovered at runtime — differ from plan): - entity.findAll() → array of {id: string, name: string} tables. - localPosition / localScale expose number components .x/.y/.z. - localRotation exposes quaternion components .x/.y/.z/.w. - entity(id).getParent() → parent entity proxy or nil; its `.id` is the id string. - entity(id).name → string name (direct field, no function call). - lighting snapshot is read directly from settings.lighting. require @builtin::assetTypes.scene.shared.saver

isSpawnerManaged(id: ?) → void

Filter: identifies player identities (entities carrying UserIdentity) + the primary camera so we don't serialize and double them on next load.

argtypedescription
id?

entityIsRemoteSynced(id: ?) → void

A synced entity that belongs to another session: one another peer owns, or one the relay delivered (other peers' players and objects, and the ghosts the relay retains, all of which carry `origin = "relay"`). It stays out of YOUR scene's dirty overlay and canonical file, which keeps the live shared state of other sessions out of your scene. A synced entity this session built that no peer has claimed (`owner() == 0`, which is how every peer holds a scene's synced entities in edit mode, where no peer announces scene content to the room) is the scene's own, as is a synced entity this peer owns and every entity that is not synced.

argtypedescription
id?

entityIsComponentProduced(id: ?) → void

An entity a component created while the engine was running that component's own lifecycle. The engine records where every entity came from at the spawn site and answers it as `origin`; `"component"` is the value it gives a spawn made from a callback the engine dispatches — `awake`, `start`, `update`, `onEnable` and their siblings — and the engine dispatches every one of those again on the next load and the next edit↔play flip. So the component puts the entity back itself, and a record of it in the scene file describes a second copy: the load restores the record AND runs the callback, the save writes both, and the scene grows by the callback's whole output every time it is opened. A component's own attaches are already read this way — `component.add` from a component's lifecycle records the same `"component"` origin, and the record gate leaves those out — so this is that rule reaching the entities beside the components. What an AUTHOR spawns through a component method keeps the origin of the call that ran it and stays the scene's.

argtypedescription
id?

entityIsLocallyReproduced(id: ?) → void

Whether THIS session's own world puts this entity back every time: the spawner owns it (a player identity, the primary camera), a component's own lifecycle created it, or it is temporary (a bundle instantiation, a runtime-only helper). All three are statements this session made about content this session has, which is why they are the ones that license taking a record out of the overlay.

argtypedescription
id?

entityIsReproduced(id: ?) → void

Whether something other than the scene file puts this entity in the world every session: this session's own world reproduces it, or it arrived replicated from another peer.

argtypedescription
id?

standsUnder(id: ?, selfVerdict: (any) → void

Returns true if `id` itself is reproduced, or stands under a temporary ancestor. Temporariness is the one verdict that reaches descendants, because the spawn that makes a temporary root makes the tree under it in the same act: a bundle instantiation drops a temporary root with non-temporary helpers underneath, and a helper written to `entities/` re-instantiates on the next load, producing duplicates that compound on every edit→play cycle. The primary camera and every player avatar / prototype clone are spawned temporary, so their machinery is covered by the same walk. The other two verdicts speak about the entity in front of them — a record the spawner puts back, a record the relay delivers — and the author's own work below one of those belongs to the scene file. Its `parent` stands when the record set provides that parent (a peer's replicated copy of a record the same file carries), and `closeUnderParent` clears it when the set does not while the excluded parent itself stands in the world, so the entity comes back at the scene root rather than beneath something the file never names. The walk stops at an ancestor that is gone: a broken parent link is nobody's ownership claim, and the entity below it stands on its own state. Its record keeps the reference — neither the file nor the world answers to it — and a load reports the record it cannot place rather than standing the entity at the scene root as one the scene has no record for.

argtypedescription
id?
selfVerdict(any

entityIsUnderReproduced(id: ?) → void

argtypedescription
id?

entityIsUnderLocallyReproduced(id: ?) → void

The same walk over the local half of the verdict: this session's own world puts the entity — or the temporary root it stands under — back every time.

argtypedescription
id?

isAuthored(id: ?, _layerName: ?) → void

Decide whether an entity belongs in the dirty file for `layerName`. Drops anything the scene file does not put in the world: entities the spawner manages, temporary ones and everything under them, and those replicated in from another peer. A nil/empty layerName treats every authored entity as in-scope.

argtypedescription
id?
_layerName?

componentShortName(t: string) → string

The component type's short name: the leaf of the identity every spelling of the component resolves to, so a saved record (`Model`, `Door`) and a live one (`@builtin::components.Model`, `doors.Door`) read alike.

argtypedescription
tstringA component type name in any form.

examples

require("@builtin::assetTypes.scene.shared.saver").componentShortName("@builtin::components.Model")

transformDiffLines(oldRec: any, newRec: any, driven: { [string]: boolean }?) →

Diff lines for an entity's transform channels (position / rotation / scale), comparing the live record against the record the scene carries. A channel the saved record OMITTED defaults to the identity transform, so a move away from origin registers even for an entity saved at identity (which carries no transform block) — while an unmoved entity whose live record materializes the identity transform produces no line. A channel named in `driven` holds the running simulation's state and takes no part. drove in play, as a set; nil compares every channel.

argtypedescription
oldRecanyThe scene's entity record (may omit `transform` or channels).
newRecanyThe live entity record.
driven{ [string]: boolean }?Channel names (`position` / `rotation` / `scale`) a component

examples

require("@builtin::assetTypes.scene.shared.saver").transformDiffLines({}, { transform = { position = { 0, 1, 0 } } })

componentDrivenFields(id: any) →

The component fields on a live entity whose current value a component callback wrote while the engine was in play, keyed by SHORT component name. A lamp component writing `Light.intensity` every frame reports `Light = { intensity = true }`; an authoring call writing that same field takes it back off. These name the running simulation's state rather than authored intent, so the record comparison passes over them. type resolves to without its library prefix; empty for an entity no component drove.

argtypedescription
idanyThe live entity's id.

examples

require("@builtin::assetTypes.scene.shared.saver").componentDrivenFields("ent_0")

transformDrivenChannels(id: any) →

The transform channels of a live entity whose current value a component callback wrote while the engine was in play, as a set. A camera rig turning its camera every frame reports `{ rotation = true }`; an authoring call writing that same channel takes it back off. These name the running simulation's state rather than authored intent, so the record comparison passes over them.

argtypedescription
idanyThe live entity's id.

examples

require("@builtin::assetTypes.scene.shared.saver").transformDrivenChannels("ent_0")

withoutDrivenState(rec: any, base: any?) → any

An entity record with the running simulation's state taken out: every transform channel and component field a component drove in play on the entity the record names carries the value `base` states for it, and is left out where `base` states none. With the scene's record as `base`, it is what a verdict writes into the scene for a live entity: accepting an author's change records that change and leaves the value a running component happened to hold that frame where the scene had it. With no `base`, it is what a write to the scene applies to a live entity in play, leaving each channel and field a running component drives to that component. read. leaves them out.

argtypedescription
recanyThe entity record; its `id` names the entity whose driven state is
baseany?The record whose values the driven channels and fields take; nil

examples

local ss = require("@builtin::assetTypes.scene.shared.saver"); ss.withoutDrivenState(ss.serializeEntity(id), saved)

recordDiffLines(oldRec: any, newRec: any, maxLines: number?) →

Human lines describing how a live entity record differs from the record the scene carries: transform channels, component fields, component additions and removals, and attribute additions, changes and removals. Compares authored intent: a component field participates where either record states it and while its live value is an author's (a field only the live record states was set after the scene's record was made, and one the instance wrote about its own runtime is left out), numbers compare within a float-noise epsilon, and asset references compare by the asset they name rather than by the shape they are written in. A field or transform channel a component drove in play holds the running simulation's state, which the scene's record puts back, so it states no difference. An EMPTY result is the "unchanged" verdict.

argtypedescription
oldRecanyThe scene's entity record.
newRecanyThe live entity record.
maxLinesnumber?How many lines to report before summarizing the rest as "…".

examples

local ss = require("@builtin::assetTypes.scene.shared.saver"); ss.recordDiffLines(saved, ss.serializeEntity(id))

push(s: string) → void

argtypedescription
sstring

sceneFolderFor(target: ?) → void

Resolve `target` (full VFS path with / without `scene.json`, or a bare scene name) to the .scene folder path with trailing `/`.

argtypedescription
target?

canonicalPathFor(target: ?) → void

argtypedescription
target?

dirtyDirFor(target: ?) → void

Dirty directory: <scene folder>/scene_dirty/. Per-entity body files live alongside the manifest under entities/.

argtypedescription
target?

dirtyManifestPathFor(target: ?) → void

argtypedescription
target?

dirtyEntitiesDirFor(target: ?) → void

argtypedescription
target?

dirtyEntityFilePathFor(target: ?, entityId: ?) → void

argtypedescription
target?
entityId?

tombstoneBody(id: ?) → void

argtypedescription
id?

isTombstone(body: ?) → void

argtypedescription
body?

withdrawnKey(target: ?) → string

argtypedescription
target?

isWithdrawn(target: ?, eid: ?) → boolean

argtypedescription
target?
eid?

noteWithdrawn(target: ?, eid: ?, wasLive: boolean) → void

argtypedescription
target?
eid?
wasLiveboolean

endWithdrawalForMark(target: ?, eid: ?) → void

End `eid`'s withdrawal in `target` because a mark named it. A mark is a statement the author made after the verdict, and the scene takes it — with one exception the withdrawal already answered: a withdrawn removal whose entity is still absent, where the mark restates that removal and the withdrawal stands.

argtypedescription
target?
eid?

clearAbsentWithdrawals(target: ?) → void

Drop `target`'s withdrawn removals. Called wherever the scene is composed: the composed body is the record a withdrawn removal was made to keep, so reading it out is that withdrawal answered.

argtypedescription
target?

recordCanonicalIds(target: ?, body: ?) → void

argtypedescription
target?
body?

wasCanonical(target: ?, eid: ?) → void

argtypedescription
target?
eid?

ledgerKey(target: ?) → string

argtypedescription
target?

noteHeld(target: ?, eid: ?) → void

Record that `eid` is live in `target` for this session.

argtypedescription
target?
eid?

isHeld(target: ?, eid: ?) → boolean

Whether this session has ever seen `eid` live in `target`.

argtypedescription
target?
eid?

noteBuildRemoved(eid: string) → void

Say that a scene build took the entity `eid` out of the world. A save of the scene then states the removal of the record the scene holds for it, however the build ran. A build that runs as its scene loads reaches `scene.json` this way and records nothing in the scene's unsaved edits.

argtypedescription
eidstringThe entity id the build despawned.

examples

require("@builtin::assetTypes.scene.shared.saver").noteBuildRemoved(id)

noteStatedRemoval(target: ?, eid: ?) → void

Record that this session stated the removal of an overlay-only `eid`.

argtypedescription
target?
eid?

resetStatedRemovals(target: ?) → void

Drop the removals this session is carrying for `target`. The records they name are what they track, so emptying the overlay is what retires them.

argtypedescription
target?

forgetHeld(target: ?, eid: ?) → void

This session stops claiming it held `eid` in `target`. What a session says about an entity's absence rests on having had it, so an entity it no longer claims is one the removal reconciliation leaves to whoever authored it, and the statements this session already made about that entity retire with the claim.

argtypedescription
target?
eid?

statedRemovalIds(target: ?) → void

The overlay-only removals `target` carries, stated and published alike.

argtypedescription
target?

add(set: { [string]: boolean }?) → void

argtypedescription
set{ [string]: boolean }?

retireUnrecorded(target: ?, removed: { string }) → void

Let go of what this session keeps for an entity no record of `target` names any more: its held id, and the handles its records left out, which only a load building the entity from a record puts back. A held id is what lets this session state a removal about a record the scene still carries; an id whose entity is gone, which the canonical file the last compose read does not carry, and which an empty overlay names nowhere, is an entity whose removal the scene already states by the row's absence, and nothing reads it again. `removed` names the overlay-only removals the emptied overlay took away, which left the held set when they were stated. Called once the overlay is empty, so what is kept is the size of the entities the scene carries rather than of every entity this session ever held in it. A scene no compose has read keeps its held set: the file may still carry any of them.

argtypedescription
target?
removed{ string }

noteRecordWaits(target: ?, eid: ?) → void

Record that `eid`'s overlay record in `target` waits for a parent no entity provides. While it waits, the saver writes nothing over that record from the entity's live state, until a mark on the entity or `endRecordWait` ends the wait.

argtypedescription
target?
eid?

endRecordWait(target: ?, eid: ?) → void

End the wait `noteRecordWaits` recorded for `eid` in `target`, so the saver records that entity's live state again.

argtypedescription
target?
eid?

recordWaits(target: ?, eid: ?) → boolean

Whether `eid`'s overlay record in `target` waits for its parent, so the saver leaves that record as it stands.

argtypedescription
target?
eid?

fireOnBeforeSave( ) → void

lightingSnapshot( ) → void

Scene-level lighting for the manifest delta. The scene.json `lighting` block persists only the non-entity state the loader's `settings.lighting` schema owns: `clear_color` and the legacy `sky` config block. Lights and the sky are entities and are captured by the entity delta like everything else. Returns nil when the settings block isn't available (early boot, headless harness) so the delta walker leaves the `lighting` key out.

defaultPlayerCfg( ) → void

defaultCameraCfg( ) → void

unwrapConfig(block: ?, fallback: ?) → void

Unwrap a typed config block (metatable-wrapped — see `newConfigBlock` in layers.module) into a plain copy of the backing data so Json.encode + the manifest delta-walker can iterate it normally. The wrapper's own table is empty (__newindex needs to fire on every write, so the wrapper can't store data on itself); without unwrapping, the saver would serialize `player = {}` regardless of what the user set. A plain-table fallback (no _snapshot method) passes through directly for callers that build a SceneProxy by hand or pre-typed-block tests.

argtypedescription
block?
fallback?

proxyForLayer(layerName: ?) → void

Resolve the scene proxy for `layerName` (guid or display name) via the live layers registry, or nil when none is found.

argtypedescription
layerName?

proxyConfigs(layerName: ?) → void

Snapshot the proxy's player + camera config blocks (under `proxy.settings.{player,camera}`). Authored intent only; falls back to "required in both modes, inherit world default" when no settings table is attached.

argtypedescription
layerName?

sceneFolderChecksum(target: string) → string

The checksum the scene folder of `target` carries now, or nil when it has none to read.

argtypedescription
targetstring

loadedVersionForLayer(layerName: ?) → void

The version a scene was loaded as, read from its canonical scene.json. Absent canonical (a brand-new scene) → CURRENT_VERSION. A v6 scene re-saves as v6; a v7 scene re-saves as v7, so the manifest never silently upgrades the version.

argtypedescription
layerName?

buildManifest(layerName: ?) → void

Build the manifest body — scene-level config ONLY. No entity list. The delta-overlay model derives entity identity from `canonical.entities[]` + `entities/` listing, so the manifest is small + writable without contending for an entity-list array. The `version` is the version the scene was loaded as (v6 stays v6), never a forced upgrade. v6 emits the declarative player{}/camera{} blocks; v7 emits `player` as the string intent it carries on the proxy (`playerIntent`) and omits the camera block, which v7 does not use.

argtypedescription
layerName?

recordFieldValue(v: any) →

The durable form of one component field value — what a record states for it. An asset reference keeps the identity a later session re-resolves it from; a live GPU resource handle names a slot in this session's GPU registry, so it is left out and named in the second return instead. Reading a live field through this is what lets a consumer of a record compare what it holds against what the record says in one form. was left out when one was.

argtypedescription
vanyA field value read off a live component.

examples

require("@builtin::assetTypes.scene.shared.saver").recordFieldValue(model.material) -- { __ref = "..." }, nil

layerProxyFor(layerKey: ?) → void

The live layer proxy answering to `layerKey`, or nil when no loaded layer does. Accepts either the layer's display name (resolved via asset.ref by Layers.find) or its guid (resolved by iterating Layers.list and matching proxy.guid); dirty marks arrive via the sync FFI tagged with SceneOwner.scene, which carries a guid in the guid-keyed scene model.

argtypedescription
layerKey?

layerIsAdditive(layerKey: ?) → boolean

Whether the layer answering to `layerKey` is additive. An additive layer is composed from its own source on every load, so it carries no dirty state and a mark naming one describes an entity no overlay is ever asked about — which is the one absence of a scene folder that is the layer's design rather than a record with nowhere to go.

argtypedescription
layerKey?

sceneLoadRunning( ) → boolean

Whether a scene load is in flight (`layers.loadInFlight`). The layer it brings in resolves to its scene folder once the engine holds a slot for it, which is frames after the entities the load restores are attributed to it.

targetForLayer(layerKey: ?) → void

Resolve a target VFS path for `layerKey` by walking the live layer proxy. Returns nil for additive layers (they don't carry dirty state) unless a build is baking one, and for layers we can't find a proxy for.

argtypedescription
layerKey?

writeOverlayRecord(filePath: string, body: any) → string

Write `body` as the overlay record at `filePath`. Returns nil when the record is on disk, else the reason it is not — the caller decides what a record it could not write means for the operation it is part of.

argtypedescription
filePathstring
bodyany

readOverlayRecord(filePath: string) → any

The body the overlay record at `filePath` currently states, or nil when there is no record there or its bytes do not decode to one. What a record says is what the flush below decides against: a tombstone already states the removal it would otherwise state again.

argtypedescription
filePathstring

removeOverlayRecord(filePath: string, durable: boolean?) → string

Delete the overlay record at `filePath`. Returns nil when the file is gone, else the reason it is still there.

argtypedescription
filePathstring
durableboolean?

retractOverlayRecord(target: ?, eid: ?, opts: ?) → string

Take `eid`'s overlay record in `target` back out of the scene: the record is deleted, the id is withdrawn so no writer re-states it, and this session stops claiming it held the entity so the removal reconciliation has nothing to say about its absence. What the scene carries for the entity is then its canonical record, whichever way the withdrawn record differed from it. Returns nil when the record is off disk, else the reason it is still there — a caller that reports the change settled while the record still stands would be reporting work the scene never took. `opts.durable` states that this retraction settles the scene's source. A verdict on a review says it, so the record comes out with play still running and the session it is running in keeps its frames; a caller that says nothing takes the terms play is on, and reads the lock's refusal as the reason the record is still there.

argtypedescription
target?
eid?
opts?

overlayRecordIsRetractable(target: string, eid: string) → boolean

Whether `eid`'s overlay record in `target` is one this session may take out. Declining to WRITE a record and DELETING one that is already there are two different acts, and only the first is safe to base on a session-relative verdict. An overlay row is a plain VFS file that any session sharing the world can have written, so a row whose entity merely arrived replicated from a peer belongs to the session that authored it: this session composes it into the scene, and never deletes it. Two conditions therefore have to hold — this session's writer holds the entity, and this session's own world is what puts the entity back (the spawner owns it, or it is temporary or stands under a temporary root).

argtypedescription
targetstring
eidstring

retractReproducedRecord(target: string, eid: string) → void

Take `eid`'s overlay record in `target` out. A save writes no record for an entity this session's world reproduces, so a record standing from before it became reproduced states content the scene never writes: the compose bakes it into the canonical file and the publication gate counts it as a pending edit. Taking it out leaves the overlay saying exactly what a save would.

argtypedescription
targetstring
eidstring

flushEntityRecords(layerName: ?, entityIds: ?, stopAt: number?) → void

Per-entity flush. For each id, decide what the dirty file should carry by examining live ECS state + prior dirty contents + the canonical-id tracker: Live + isAuthored → write full body as override. Live + !isAuthored → RETRACT. The mark fired (Rust doesn't know the authored filter) but the entity is spawner-managed / temporary / remote; nothing to record, and any record standing from before it became reproduced comes out. Keeps `entities/` saying what a save writes. !Live + had body in dirty → write tombstone (user authored then deleted in the same session). The record is the only place the removal is stated, and every session sharing the scene reads it there. !Live + canonical-tracked → write tombstone (user deleted a or in canonical now previously-saved entity that was never touched this session, or one a verdict baked and swept from the overlay, so no dirty body file exists for the merger to find). !Live + neither → SKIP. Spawner-machinery transient (e.g. player-identity respawn on edit↔play transitions); orphan tombstones would just clutter the directory. The marks it drains are what the engine noticed; `forceRecordRemovals` is what the scene's own records say, and the two together are the whole reconciliation. `stopAt` is an `os.clock()` reading the flush stops at: once it has passed, the flush ends after the id it is on. The second return is how many of `entityIds`, from the front, the flush has settled; the rest are the caller's to hand back later.

argtypedescription
layerName?
entityIds?
stopAtnumber?

canonicalFileCarries(eid: string) → boolean

argtypedescription
eidstring

flushEntitiesForLayer(layerName: ?, entityIds: ?) → void

argtypedescription
layerName?
entityIds?

sceneRemoved(layerName: string) → void

Stop writing anything for the loaded layer `layerName`: the scene it was loaded from was removed from the world, so its dirty overlay has no scene left to belong to. The marks it holds and every mark it takes until it loads again are dropped, and nothing is written into the removed scene's folder, which would otherwise bring that folder back.

argtypedescription
layerNamestringThe layer's name, as `layers.list()` names it.

examples

Saver.sceneRemoved(proxy.name)

sceneReturned(layerName: string) → void

Write the loaded layer `layerName`'s edits again: a scene of that name loaded, so the layer has a scene to write into. Undoes `M.sceneRemoved`.

argtypedescription
layerNamestringThe layer's name, as `layers.list()` names it.

examples

Saver.sceneReturned(proxy.name)

recordsWritten( ) → number

How many entity records this session has written into a scene's dirty overlay, counted as each one lands. A record already holding the bytes it would be written with is left as it is and not counted. Read twice, the difference is the record writing done between the two reads, which is how a frame's share of a save or of the continuous writer is seen.

examples

local before = Saver.recordsWritten()

marksWaitForLoad(layerKey: ?) → boolean

Whether the writer holds the marks for `layerKey` until a scene load is through: the layer resolves to no scene folder while a load is in flight, which is how the layer a load brings in reads until the load hands it back, and a build running inside the load marks what it changes in the scene before then. An additive layer's marks are discarded by the flush whenever they reach it.

argtypedescription
layerKey?

releaseParked( ) → void

Put back into the backlog the marks that waited for a scene load, for each layer that resolves now, and for every layer once no load is in flight: a layer that still resolves nowhere then is one no load is bringing in, and the flush answers for its marks. A mark that waited is taken the way `takeMarks` takes one whose layer resolves, so it ends a withdrawal and a record's wait for its parent the same.

takeMarks( ) → void

Take the engine's dirty marks into the backlog. A mark is an author acting on the entity after any verdict withdrew its record, so the withdrawal is over and the record written for it says what the mark says: the body of an entity that is here, the tombstone of one that has since been removed. One mark leaves the withdrawal standing: an id still absent, restating the removal the verdict took back.

settleEntries(token: { thread: thread }, limit: number?, stopAt: number?) → void

The loop behind `settleBacklog`: writes the oldest waiting records until `limit` of them are written or `os.clock()` passes `stopAt`. `token` is the settle's own entry in `BACKLOG.settling`; a settle that finds another has taken that over leaves the backlog to it, with the records it just wrote standing.

argtypedescription
token{ thread: thread }
limitnumber?
stopAtnumber?

settleIsRunning(held: any) → boolean

Whether the settle `held` names is one this call is inside: it runs on this thread and its loop is on this thread's stack. Any other settle entry is one the caller takes over. The watchdog ends a script that overruns its budget wherever it stands, a settle included, and what that leaves is an entry whose loop never runs again.

argtypedescription
heldany

settleBacklog(limit: number?, stopAt: number?) → boolean

Write the records of the oldest waiting entries: at most `limit` of them, stopping once `os.clock()` passes `stopAt`. Each nil means no bound, which settles the whole backlog. At least one entry is settled whenever any is waiting, so a backlog always drains. Returns whether entries still wait. One settle runs at a time. A read of the overlay made from inside a record's serialization finds the settle already running and returns to it, rather than moving the head out from under the run the outer settle is part way through. A settle entry left by a script the watchdog ended is taken over, and `settleEntries` checks its token after every run it writes, so a settle that was taken over stops without moving the head.

argtypedescription
limitnumber?
stopAtnumber?

settleWaitingRecords( ) → void

Write every waiting record now. The overlay a reader lists is then every mark the writer has taken.

deepEqual(a: ?, b: ?) → void

Structural equality. Two nil values are equal; two tables are equal if every key in `a` is in `b` with a deepEqual value, and `b` has no extra keys.

argtypedescription
a?
b?

flushManifestForLayer(layerName: ?) → void

`flushManifestForLayer` below reaches for `readCanonical`, defined further down with `composeMerged` (they share helpers) and bound to the forward declaration near the top of this module. Rewrite the manifest for `layerName` as a TRUE DELTA: only the scene-level fields that differ from canonical land in the overlay. If the live state matches canonical (or no canonical exists yet and the live state matches the saver's defaults), nothing is written and any stale manifest is removed. The merger in `composeMerged` already treats missing overlay fields as "use canonical's value", so writing only the diff is sufficient. Same property as `entities/`: edit nothing → no overlay file created; edit one lighting field → only the `lighting` block appears in the overlay. Without this, every spawner-machinery mark would flood the manifest with full default snapshots and make it impossible to tell what the user actually changed. Returns one line per manifest state this pass could not put on disk. The manifest carries the scene's own lighting / player / camera config, so a delta it cannot write is authored work the composed scene does not carry and a stale block it cannot delete is config the next load applies over the author's own.

argtypedescription
layerName?

liveAuthoredIds(layerKey: ?) → void

The live entities of the ACTIVE layer, which the explicit persist points (save / writeDirty) re-serialize. The continuous writer only re-serializes entities whose per-entity dirty mark has landed, and a public-field value edit reaches the ECS without landing such a mark, so the incremental overlay keeps the entity's add-time body; reading the live set at a persist point captures current ECS truth regardless of which marks fired. The per-edit continuous-writer cost stays O(1); that O(authored) pass runs only on an explicit save or a mode flip. Keyed to the active layer, which is what `serializer.activeLayerEntityIds` enumerates; `layerKey` is the guid (save) or name (writeDirty) the overlay was written under, so `flushEntitiesForLayer` resolves the same path.

argtypedescription
layerKey?

recordAcrossFrames(layerKey: ?) → void

The record pass of a save, spread over frames at the continuous writer's rate: the marks already taken, then every live authored entity of the layer, at most `FRAME_RECORD_BUDGET` records a frame, a frame boundary between one share and the next. A save runs in a scheduler task, so how long its frames take is set by that budget rather than by how many entities the scene holds. The marks that land while it runs go into the backlog, and the pass ends by settling what they left, so the overlay it hands the compose is every mark taken up to its last frame and a record of every entity live in it. Returns one line per record it could not put on disk, as the force-flush does.

argtypedescription
layerKey?

isCodeReproduced(eid: string) → boolean

Record every entity live in the active layer as one this session holds, so the removal reconciliation can tell an entity this session had and lost from one it never built. A load is the moment a scene's whole set becomes live at once, and it happens before the session's first persist point, which is why the load calls this rather than waiting for the flush to sample it. An entity the composed scene names is noted whatever made it: the record is what a removal is stated against. An entity no record names and that code puts back on every load is left out: anything an entrypoint or component created under a fresh id (the freeze gate's reproduced contexts), the spawner's player body and camera, a component's own output, a temporary tree. Each load makes those under new ids, and a ledger that noted them would gain ids per load that nothing reads. The record lookup comes first and the freeze gate's cached tag second, so a scene the load built from its records pays one table read per entity.

argtypedescription
eidstring

noteHeldEntities(layerKey: ?) → void

argtypedescription
layerKey?

readDirtyEntityOverlay(target: ?) → void

Read every file under `entities/` and return a list of { id, body, tombstoned } rows. Files that fail to decode are dropped with a warning so a single bad file doesn't break the load. A file that exists and whose bytes this engine cannot read raises, naming it. The filesystem listing IS the source of truth for the dirty entity set — the manifest deliberately doesn't carry an id list, so the composer never has to reconcile a stale list with the actual files on disk.

argtypedescription
target?

reconcileReproducedRecords(target: string) → void

Sweep the overlay for records of this session's own that the live world has since made reproduced, so a compose reads only what a save would write. The per-mark retraction above covers every entity the engine marks; this covers the one it does not mark again after the change — a spawn marked temporary by the code that made it — and runs at the persist points, which already walk the overlay. A row for an entity replicated in from a peer is left exactly as it stands: the session that authored it says what the scene carries for it.

argtypedescription
targetstring

decodeCanonicalBody(path: ?, src: ?) → void

Read the canonical scene.json into a body table. Returns nil when canonical doesn't exist; falls back to an empty-scene body on decode failure so a bad canonical doesn't lock the user out. Bound to the forward-declared upvalue above so `flushManifestForLayer` (defined earlier) sees this function. Decode one canonical body, or an empty scene when the text does not parse. Shared by the fresh read and the cached one so both report a bad file the same way.

argtypedescription
path?
src?

contentAddressOf(path: string) → string

The address the bytes at `path` are stored under, or nil where memory holds them (a synchronous read returns those) or nothing does.

argtypedescription
pathstring

holdCanonicalText(path: string, text: string, address: string?) → void

Take `text` as what `path` holds, under `address`: the one the path answered before the read that produced `text`. A read that yields can finish after a write has replaced the bytes, so the copy is taken only while the path still answers that address. Text memory holds needs no copy: a synchronous read returns it.

argtypedescription
pathstring
textstring
addressstring?

watchCanonicalText(path: string) → void

Watch a canonical file this engine read, so the text each write lands is taken while it is still in reach. A file read while small is watched too: a later write can carry it past the size memory keeps, and that write is the moment its text is readable without a fetch.

argtypedescription
pathstring

heldCanonicalText(path: string) → string

The held text of `path`, while the address it was taken under still names the bytes the path holds.

argtypedescription
pathstring

readInReach(path: string) → void

The bytes at `path` as this caller can read them, or nil and the reason the read found none. Code that can yield waits for bytes still on their way, and a fetch that finds none answers nil and its reason. Code that cannot yield is refused with `VFS_READ_CANNOT_WAIT` for them, and that refusal answers nil and the refusal: the file exists and is out of this caller's reach, which the held copy answers for. Any other refusal is raised as it came.

argtypedescription
pathstring

readCanonicalSource(name: ?) → void

Read the canonical source text, or nil when there is nothing to read. The third answer is set when the file exists and this caller cannot read it: the read found no bytes and no held copy answers for them, while the path is still there. It names why: the caller's context when the read refused it bytes still on their way (the caller runs where a coroutine cannot yield, and a caller that can yield waits for them), and the read's own reason otherwise (a fetch that found no bytes, which every caller meets alike). That is a different answer from an absent file, and each reader decides what it means for what it reads.

argtypedescription
name?

readCanonical(name: ?) → void

The canonical body for `name`, a table of the caller's own, or nil when the scene has no canonical file. Every reader of the whole body composes, diffs or rewrites the scene from it, and a body read as absent while the file exists is a scene with every entity it carries missing: a load would place the overlay alone, the leave-play count would name each of the file's entities as play work, and a save or a verdict would write the scene without them. A file this caller cannot read therefore raises, naming the file and why.

argtypedescription
name?

readCanonicalShared(name: ?) → void

The canonical body for `name`, decoded once per distinct file content. The returned table is shared with every other holder of the same content, so it belongs to readers. `readCanonical` returns a body of the caller's own for anything that edits what it reads — `composeMerged` folds the dirty overlay into its copy in place. A dirty flush reads the canonical body twice per frame and looks at scene-level fields only (`version` for the manifest's version, and `lighting` / `player` / `camera` for its delta). Decoding a saved scene to reach them costs the whole document, entity list included, so on a large scene a per-frame transform edit spent its frame in `Json.decode`. A file this caller cannot read answers nil, as an absent one does, and the reason as a second answer, which an absent file does not give. The continuous writer reads through here from the engine's frame, where a coroutine cannot yield, for the scene's version and config, which fall back to the current version and the saver's defaults; a raise there would drop the records of that frame instead. Those records go to the overlay, never to the file, so the file's own records stand.

argtypedescription
name?

canonicalEntityIdsFor(name: ?) → void

The entity ids canonical names for `name` — what `scene.json` carries on its own, before the dirty overlay merges over it. The removal reconciliation asks this to pick the record it writes for an absent entity: an id canonical carries needs a tombstone over it, because a plain absence from the overlay lets the merge carry the canonical row through. The returned table is shared with every other caller reading the same canonical content, so it belongs to readers.

argtypedescription
name?

namesLiveEntity(ref: string) → boolean

Whether `ref` — a record's `parent`, in either spelling a load resolves one against — names an entity standing in the world right now. `entity.findAll` answers for an id and for a display name in the one call. The two answers cost different things, so a probe that cannot be made answers the cheaper one: "the world provides it" clears the reference and composes the record at the scene root, which keeps the entity across the round trip, while "nothing provides it" leaves a reference the load abandons the record over. A world that cannot be asked is not evidence that a record is broken.

argtypedescription
refstring

closeUnderParent(name: ?, entities: ?, aside: { [string]: boolean }?, forWrite: boolean?) → void

Close a composed record list under `parent`: every `parent` a record carries names a record the SAME list carries, by id or by display name — the two spellings a load resolves a parent against. A record whose parent names neither is answered by what the WORLD says about that parent, and the two answers are different repairs: * The parent stands in the world and the file simply leaves it out — a temporary root, a spawner's own machinery, the entity a component rebuilds. The author's work below it belongs to the scene, so the reference is cleared and the record composes at the scene root: the file is closed under `parent` and the entity survives the round trip. * Nothing answers to the reference at all. Clearing it would turn a broken record into a root entity the author never made, which is what a load then puts in the world and a save then writes down as authored. The reference stands instead, so the record keeps stating the parent it wants and a load reports the one record it cannot place rather than inventing a placement for it. Both are reported once, naming the scene and a sample of the records. A composition for a write (`forWrite`) reports only the first: the records of the second kind are held back from the file it writes (`splitUnplaced`), and the writer reports them as it holds them back. `aside` names, by id and by display name, the records a save keeps in the overlay beside the list it writes. They are records the scene carries, so a reference to one stands, the same as a reference to a record in the list. The rule lives here because this is the one list both ends read. The canonical bytes a save writes and the records a load spawns are the same composition, so a file left behind by an older engine, a hand edit, or a partial sync is answered on the way in as well as on the way out.

argtypedescription
name?
entities?
aside{ [string]: boolean }?
forWriteboolean?

splitUnplaced(entities: { any }) → void

Split a composed record list into the records a load of that list alone places and the records it abandons: a record whose `parent` no record of the list provides, by id or by display name, and every record beneath one, whatever order the records come in. The first return keeps the list's order and the second holds the abandoned records in the order they had in it. One pass: each reference counts the records of the list that provide it, and a record that is held back takes its references out of those counts, so the records under a reference no record provides any longer are held back in turn. A display name several records carry stays provided while any of them is kept.

argtypedescription
entities{ any }

refsOf(rec: any) → void

The references a record provides: its id, and its display name beside it.

argtypedescription
recany

buildIdentityOf(rec: any) → void

The build a record is the output of: the `buildOwner` and `buildKey` attributes a build's reconcile writes onto every entity it places, which a record carries beside the entity's name. Nil for a record no build placed.

argtypedescription
recany

buildOutputIndex(records: { any }) → void

The record id each build output among `records` stands for, by build and record identity.

argtypedescription
records{ any }

sceneBuildModule( ) → any

The scene build module, or nil where it cannot be loaded.

replacedByItsBuild(rec: any, written: { [string]: string }, SceneBuild: any) → boolean

Whether `rec`, a record no load of the records `written` beside it places, is build output its build has replaced. Its entity stands nowhere in the world, and either `written` carries another record for the same build and record identity (the build placed that identity again, on a new entity), or the build ran in this session and places nothing for that identity any more (`SceneBuild.states`). A build states its whole output on every run, so such a record states nothing the scene carries.

argtypedescription
recany
written{ [string]: string }
SceneBuildany

splitReplacedBuildOutput(written: { any }, held: { any }) → void

Split the records a write holds back (`splitUnplaced`) into the ones that wait for their parent and the build output its build has replaced (`replacedByItsBuild`), which the write takes out of the scene. `written` is the record list the write carries. Both returns keep the order `held` had.

argtypedescription
written{ any }
held{ any }

recordsAwaitingParent(target: string, overlay: { any }?) → void

The records of `target` that wait for their parent: the ids of the records a load of the scene abandons because the parent each names is provided by no record it places and by no entity standing in the world. A record under one that waits waits with it, whatever order the records are in. `overlay` is the overlay's rows when the caller already read them. Such a record states an entity that stands nowhere yet, and goes on stating it until an entity providing its parent arrives, so its entity's absence is the wait. A save leaves the record in the overlay, where the overlay applier builds it once its parent builds; every load of `scene.json` abandons it. One pass over the records: each reference counts the placed records that provide it, by id and by display name, and a record that starts to wait takes its references out of those counts, so the records under a reference no placed record provides any longer wait in turn.

argtypedescription
targetstring
overlay{ any }?

standsInWorld(ref: string) → boolean

argtypedescription
refstring

nameBeside(key: string, rec: any) → string

The display name `rec` provides beside its key, or nil when it provides none or the name is its key.

argtypedescription
keystring
recany

wait(key: string) → void

argtypedescription
keystring

release(ref: string) → void

A reference whose last placed provider now waits: the records under it wait with it, as a load abandons every record under one it abandons, whatever entity the world holds under that id or name. The world answers only for a reference no record provides at all.

argtypedescription
refstring

readDirtyManifest(name: string) → any

The dirty manifest for `name` as a table, or nil when there is none to read.

argtypedescription
namestring

mergedSceneConfig(canonical: any, manifest: any) → void

The scene-level config a bake of `manifest` over `canonical` writes. A scene's v6/v7 identity is fixed by its CANONICAL body, never by a dirty overlay. Genuinely v7 = canonical version 7, or an authored string `player` intent. A legacy v6 scene (declarative player{}/camera{} blocks, numeric version below 7) stays v6: a stray `version: 7` or string `player` overlay — a defaulted "spawns" intent the v7 machinery persisted onto a legacy scene — does not upgrade it, or the loader discards its procedural player/camera and errors on every entity for a missing `synced` flag the v6 records never carried. A genuinely-v7 scene carrying a string `player` intent then settles its version to 7 and drops any legacy `camera` block a prior v6 serialization left behind, so a v7 scene stuck at version 6 heals to a clean v7 body on the next save instead of perpetuating the mismatch.

argtypedescription
canonicalany
manifestany

overlayEmptying(name: ?, canonical: ?, rows: ?) → void

Whether `rows`, the dirty overlay of the scene `name`, carries a tombstone for every entity the canonical body holds: `{ entities, tombstones, path }` when it does, nil when the overlay leaves the scene something.

argtypedescription
name?
canonical?
rows?

composedSceneBody(name: ?, opts: { setAside: { [string]: boolean }?, keepSavedWhenEmptied: boolean? }?) → void

Canonical + dirty overlay composed into one body, by the rules `composeBody` states: the body, and the overlay `opts.keepSavedWhenEmptied` left out of it.

argtypedescription
name?
opts{ setAside: { [string]: boolean }?, keepSavedWhenEmptied: boolean? }?

composeMerged(name: ?, opts: { setAside: { [string]: boolean }?, keepSavedWhenEmptied: boolean? }?) → void

The composed body, with the canonical-id tracker refreshed to the entity set it states. Composing for a load settles what the scene holds, so the tracker moves with it: a subsequent despawn mark reads the merged view to decide whether it needs a tombstone (the entity was in the scene) or nothing (it was transient spawner machinery). A reader that only wants to know what the scene composes to calls `composedSceneBody` and leaves the tracker where the last load put it.

argtypedescription
name?
opts{ setAside: { [string]: boolean }?, keepSavedWhenEmptied: boolean? }?

composeForWrite(name: string, setAside: { [string]: boolean }) → void

The body a write of scene `name` puts in its `scene.json`, with the records `setAside` names left in the overlay: `composeBody` with `closed`, the canonical-id tracker refreshed as `composeMerged` refreshes it. Returns the body, the records held back to wait for their parent, and the build output its build has replaced.

argtypedescription
namestring
setAside{ [string]: boolean }

sceneBodyForReading(name: string) → any

The body a scene composes to, for a caller that only reads it. A scene with no dirty state composes to its saved `scene.json`, so the body is the one decoded for the file's current bytes and shared with every other reader of them: asking again costs no decode while the file holds still. A scene with dirty state is composed afresh, as `composedSceneBody` does. The tracker of canonical ids is left where the last load put it. The table belongs to its readers and is never written to; `composeMerged` is the call for a body the caller edits. A saved file that exists and that this engine cannot read (its bytes are not on this engine, or the caller cannot wait for them) raises, naming the file and why, as `composeMerged` does.

argtypedescription
namestringThe scene: its path or identity.

examples

local body = Saver.sceneBodyForReading("/source/scenes/arena.scene"); print(#body.entities)

overlayEmptiesScene(name: string) →

Whether the scene's dirty overlay carries a tombstone for every entity its saved `scene.json` holds, which composes the scene to nothing it was saved with. An overlay of a scene that is being edited removes some of its entities; one that removes all of them is a record of something other than an author's edits (a build that stated nothing, a damaged journal), so the loader loads the saved scene instead and the pending-edit verdict names it. saved scene holds, how many of them the overlay removes, and the overlay folder.

argtypedescription
namestringThe scene: its path or identity.

examples

local emptied = Saver.overlayEmptiesScene("/source/scenes/arena.scene")

buildOutputOf(savedRec: any?, pendingRec: any?) → string

The scene build whose output a pending edit is, or nil when an author made it. A scene build (a scene's `build.luau`, a `SceneModule` placement) writes its name, the record each entity stands for and what it stated onto every entity it places, and a rebuild that states something new restates that last part; those attributes travel in the records, so the answer is read off the two records alone. An edit is the build's output when the build placed the entity, the pending record holds exactly what the build stated (`SceneBuild.recordDrift` finds nothing an author changed or added), and, for an update, the build's statement is not the one the saved record holds. A removal is the build's when the build that placed the entity has run in this session and no longer states it (`SceneBuild.states`); an entity its build still states was taken away by an author, and the build's next run puts it back. an addition. removal. that build has run in this session (`/game/district.sceneModule/init.luau`), else by its placement (`placement 'districtHost' (ent_…)`) or its name; nil for an author's edit.

argtypedescription
savedRecany?The record the scene's saved file holds for the entity, nil for
pendingRecany?The record the scene's unsaved edits hold for it, nil for a

examples

local by = require("@builtin::assetTypes.scene.shared.saver").buildOutputOf(saved, pending)

attributesOf(rec: any?) → void

argtypedescription
recany?

pendingConfigKeys(name: string, canonical: any) → void

The scene-level config keys a bake would write differently from the way the canonical file states them. Read through `mergedSceneConfig`, the same decision `composeMerged` writes by, so a key the merge declines to apply — a `version: 7` or a string `player` standing over a legacy v6 canonical — is not reported as pending over a bake that provably leaves the file alone.

argtypedescription
namestring
canonicalany

pendingRecordDiffLines(oldRec: any, newRec: any) → void

What baking one overlay record over the record the scene carries would write, as human lines — AND the verdict that it would write anything: an empty result is the "the canonical file already says this" verdict. One walk answers both, over the same `diffEqual` the merge's own equality uses. Every top-level key the two records disagree on is reported, by the field-by-field renderers where they speak about it and by name where they do not — so a record the gate refuses over always names what it refuses over, and a record the gate passes has nothing left unsaid.

argtypedescription
oldRecany
newRecany

pusher(key: string) → void

argtypedescription
keystring

trackingReason( ) → string

Why an authored scene edit made from THIS call's own context would not be recorded right now, or nil while it would be. A scene records its edits in edit mode, outside a scene load, outside the world entrypoint's respawn pass, and outside a component or entrypoint callback tick; inside any of those a spawn or a change reaches the live session and no file. This is the question an empty `pendingSceneEdits` list cannot answer on its own — the same list stands for "the scene is saved" and for "nothing is watching it".

examples

require("@builtin::assetTypes.scene.shared.saver").trackingReason()

pendingSceneEdits(name: string) →

Every edit a scene's dirty overlay holds that baking it would write into the canonical scene file: an added entity, an updated one, a removed one, or a change to the scene's own lighting / player / camera config. An EMPTY result is the verdict that the canonical scene already says what the overlay says, whatever rows the overlay carries. Each entry names the entity and describes the difference. scene name. `"updated"`, `"removed"` or `"config"`, and `builtBy` names the scene build whose output the edit is (`buildOutputOf`), nil for an author's edit. Second, `trackingReason()`: why an authored scene edit made from THIS call's own context would not be recorded right now, or nil while one would be. It describes the caller's context, so it answers the same way for a scene the list is complete for.

argtypedescription
namestringThe scene — a `.scene` folder path, a `scene.json` path, or a bare

examples

require("@builtin::assetTypes.scene.shared.saver").pendingSceneEdits("/zero/source/scenes/main.scene")

hasPendingSceneEdits(name: string) → boolean

Whether baking the scene's dirty overlay into its canonical file would change the scene. False when the overlay is absent, and false when it holds only records the canonical file already states. scene name.

argtypedescription
namestringThe scene — a `.scene` folder path, a `scene.json` path, or a bare

examples

require("@builtin::assetTypes.scene.shared.saver").hasPendingSceneEdits(layers.active.name)

standingConfigKeys(name: string) → void

The config keys the scene still states a difference over, read through the verdict itself so a discard and the gate cannot answer differently. Nil when the scene has no config difference left.

argtypedescription
namestring

discardOverlayManifest(name: string) → string

Take the scene's overlay manifest out, so the scene's config is the one its canonical file states. A difference the canonical body carries on its own — the version and camera keys a legacy scene settles the next time it is written — outlives the manifest, so the removal is followed by the verdict's own question and a difference still standing is said as the reason rather than reported as dropped.

argtypedescription
namestring

discardSceneEdits(name: string, select: (string | { string }) →

Take a scene's unbaked overlay edits back out, leaving the canonical file as what the scene carries. This is the second verdict on the edits `pendingSceneEdits` reports and a publication call refuses over: staging bakes them into the scene, this drops them. Each edit's overlay record is deleted and its entity withdrawn so no writer re-states it, whichever session wrote the record and however long it has stood. scene name. them. Omitted, every edit the scene has pending. and the ones still standing, each with the reason it does. While play runs unpaused the lock holds every removal under `/source`, so the edits come back blocked carrying that reason; `engine.paused = true` or `engine.mode = "edit"` opens the overlay.

argtypedescription
namestringThe scene — a `.scene` folder path, a `scene.json` path, or a bare
select(string | { string }Which edits to drop: an entity id or entity name, or a list of

examples

require("@builtin::assetTypes.scene.shared.saver").discardSceneEdits(layers.active.name)
require("@builtin::assetTypes.scene.shared.saver").discardSceneEdits("/zero/source/scenes/main.scene", "look")

forceRecordRemovals(layerKey: ?, target: ?) → void

Record, at an explicit persist point, every removal the overlay does not already carry: an entity the scene HAS a record for that is no longer live. The overlay learns of a despawn from that entity's dirty mark, and a mark is dropped wherever the engine REPRODUCES content rather than authors it — the spawn/despawn traffic of a scene load, a suppressed-marks window, a callback tick. An entity ADDED inside one of those windows still reaches the record, because `recordLiveAuthored` reads the whole live authored set at this same point whichever marks fired. This walks the other side of the same reconciliation — the records, against liveness — so an addition and a removal are equally durable however each was made. Scoped to the ACTIVE layer in EDIT mode: those are the conditions under which a record's entity being absent from the world means an author removed it. For any other layer the entities were never spawned, and the overlay takes authored marks only in edit, so absence elsewhere says nothing about intent. Scoped further to the ids this session has HELD. A scene is edited by several sessions at once, and a session that never built the entity a record names reads the same absence as the author of a despawn does — a record delivered while the session was away, an id whose entity its load did not build. What separates the two is whether the entity was ever here, so that is what is asked, and an id this session never held is left to the session whose entity it is. An entity whose record waits for its parent (`recordsAwaitingParent`) is absent because a load abandoned that record, so it is left to wait: its record stays as it stands, this session lets go of its claim on the entity, and the entity is built from the record once its parent arrives. `target` is the scene folder the caller goes on to compose from, passed in rather than re-resolved here so the removals land in the very overlay that compose reads. Returns one line per removal it could not record; empty when every removal the scene needs is on disk.

argtypedescription
layerKey?
target?

record(eid: string, inCanonical: boolean) → void

A removal is a tombstone naming the entity, whichever side of the merge the record it removes came from: canonical carries its row through unless the overlay tombstones it, and an overlay-only body is read by every other session as the entity being there.

argtypedescription
eidstring
inCanonicalboolean

recordBakedRemovals(target: ?) → void

The rows of `target`'s file whose entity no longer lives, recorded as removals: what the bake of an additive layer writes, since that layer holds exactly the entities its file names.

argtypedescription
target?

collectStatedRemovals(target: ?) → void

Take out of `target`'s overlay the removals this session stated a generation ago. An overlay-only id is absent from canonical, so a load composes without it whether or not its tombstone is still on disk — what the record does after the compose is reach the sessions already holding the entity, and one persist point of grace is what it is kept for. The overlay is then back to the size of the live authored set, so a scene rebuilt many times in one session (despawn- all + respawn with fresh ids) does not accumulate a file per entity ever made. Only this session's own statements are collected, and only while the record still says what this session wrote: a body at that path is another session authoring an entity onto the id, and a canonical row makes the tombstone the durable thing that suppresses it.

argtypedescription
target?

unrecordedRecordsError(op: string, name: any, unrecorded: { string }) → string

The message a persist point refuses with when it holds records it cannot put on disk. Composing a scene over an overlay missing them writes a file that still holds entities the author removed and lacks the ones they authored, so the operation states what it could not do instead of reporting the write it went on to make.

argtypedescription
opstring
nameany
unrecorded{ string }

joinUnrecorded(a: { string }, b: { string }) → void

Every line two record passes could not write, in one list.

argtypedescription
a{ string }
b{ string }

withoutBuildBookkeeping(rec: any, bookkeeping: { string }) → any

`rec` without the attributes a scene build writes about itself, which describe the build rather than the scene.

argtypedescription
recany
bookkeeping{ string }

recordLiveAuthored(layerName: ?, target: ?) → void

Record, at a persist point, every live authored entity of `layerName` whose record tells the scene something: one the overlay already holds a record for, one the canonical file has no record for, and one whose live record differs from the canonical one by the verdict `pendingSceneEdits` gives, leaving out the attributes a scene build writes about itself. An entity the canonical file already states as it stands is left without a record, so a persist point that finds no edit leaves the overlay as it was. Returns one line per body this pass could not put on disk. A persist point composes the scene over the overlay this pass just wrote, so a record it could not write is authored work the composed scene does not carry, and the caller states it rather than reporting the write it went on to make.

argtypedescription
layerName?
target?

writeDirty(name: ?, opts: ?) → void

Force a synchronous drain + manifest rewrite so the dirty directory reflects every pending edit-mode mutation + the current scene-level config. Used by `wld.play()` before flipping mode and by explicit `SceneProxy:writeDirty()` callers. In the delta-overlay model `writeDirty` no longer snapshots the whole scene — it just ensures every pending mark has landed on disk and the manifest carries the latest scene-level config. Cheap regardless of scene size.

argtypedescription
name?
opts?

hasDirty(name: ?) → void

True iff any dirty state exists for this target — manifest OR any per-entity overlay file. Either signals a pending change the loader needs to merge over canonical.

argtypedescription
name?

canonicalSignature(target: ?) → void

Signature of the CANONICAL scene.json bytes for `target`. Two calls return an identical string iff the on-disk canonical scene is byte-for- byte unchanged. The dirty overlay is deliberately excluded: it is the live session's working state and the continuous dirty writer rewrites it on every authored edit, so folding it in would make an idempotent same-scene reload trigger a full respawn on every keystroke. The idempotent same-guid `layers.load` uses this to tell a no-op (canonical unchanged) from a reload (canonical edited via write_file / edit_file). Returns "" when the canonical file is unreadable (treated as "unknown", which the caller resolves by reloading).

argtypedescription
target?

clearDirtyPerFile(name: ?, keep: { [string]: boolean }?) → void

Remove every file under the dirty directory one at a time. Fallback for environments where recursive remove is disabled (`vfs.allow_recursive_remove = false`).

argtypedescription
name?
keep{ [string]: boolean }?

overlayHoldsOnly(name: ?, keep: { [string]: boolean }) → boolean

Whether the dirty overlay of `name` holds no manifest and no entity record but those of the entity ids `keep` names.

argtypedescription
name?
keep{ [string]: boolean }

clearDirty(name: ?, keep: { [string]: boolean }?) → void

Wipe the dirty overlay (manifest + every per-entity file), all but the records of the entity ids `keep` names, which stay where they are. Returns true if anything was removed. The per-frame writer can accumulate thousands of per-entity files across one edit session (every map rebuild despawns the previous entities and spawns fresh-id ones). Removing them one VFS call at a time is an FFI round-trip per file — a bake on a large overlay then runs for minutes. A single recursive remove collects the subtree once and drops it in one operation, so clearDirty cost is independent of overlay size.

argtypedescription
name?
keep{ [string]: boolean }?

filesSignature(target: ?) → string

What a scene's files hold, as one string: its canonical `scene.json` and every file of its dirty overlay, each with its bytes. Two reads give the same string exactly when neither the canonical file nor any overlay file changed between them, which is how a caller that rewrites files underneath a loaded scene tells whether it moved that scene. `scene.json` path, or a bare scene name.

argtypedescription
target?The scene's authoring target: a `.scene` folder path, its

examples

local before = require("@builtin::assetTypes.scene.shared.saver").filesSignature(layers.active.asset.path)

addFile(path: string) → void

argtypedescription
pathstring

respawnFromFiles(layer: any, target: string, change: (() →

Take a loaded scene's live layer to what its files hold: the layer is respawned from the canonical `scene.json` composed with the overlay records standing beside it, with the scene-load gate held so the entities that come back are not marked as edits. `change`, when given, runs first inside the same gate: it is the act that changes the files, such as a discard deleting the overlay. Once the overlay holds nothing, what this session keeps about the overlay's records is let go, as clearing the overlay lets it go. Returns once the respawn's detached work (player, camera, avatar) has settled. Raises when the respawn raises.

argtypedescription
layeranyThe layer holding the scene (a `SceneProxy` from `layers.list()`).
targetstringThe scene's authoring target, the path its overlay is filed under.
change((Optional function run inside the gate before the respawn.

examples

local Saver = require("@builtin::assetTypes.scene.shared.saver")
Saver.respawnFromFiles(layers.active, layers.active.asset.path)

promoteDirty(name: ?) → void

Compose canonical + dirty overlay → write canonical → clear dirty. Idempotent; nil when there are no dirty deltas to promote. A canonical write the VFS declines raises, the way `M.save` answers the same refusal: the clear below takes the overlay away, so that write is the whole bake, and nil is already this call's answer for a scene that had nothing to bake.

argtypedescription
name?

saveInto(name: ?, opts: ?) → void

Public `layers.active:save(opts)` entry point. Drains pending marks + freshens the manifest so live ECS state is captured even when the per-frame writer hasn't run since the last edit, then composes canonical + overlay → writes canonical → clears dirty. Same merge path as `promoteDirty`; this just guarantees the live state is in the overlay before composing.

argtypedescription
name?
opts?

save(name: ?, opts: ?) →

Write the live layer of scene `name` into its canonical `scene.json`: the overlay of the session's edits merged over what the file holds. A record waiting for a parent that no record and no entity provides stays in the overlay, where it is built once its parent arrives, so `scene.json` carries no record a load of it cannot place; such a record that is a build's output, when the build has since placed its record identity on a new entity or places it no longer, is taken out of the scene instead. `opts.layer` names the source layer (default: `name`), `opts.sourceGuid` and `opts.sourcePath` the scene it was loaded from. `opts.bake = true` is a build baking what it placed: an additive layer, which records no edits of its own, is then recorded into the scene's own folder for the save.

argtypedescription
name?The scene to write: its path or identity.
opts?`{ layer?, sourceGuid?, sourcePath?, bake? }`.

examples

saver.save("/source/scenes/arena.scene", { layer = "arena" })

drainManifests( ) → void

Rewrite the manifest of every layer whose scene-level config (lighting, config blocks) changed since the last drain.

drainOnce( ) → void

Take `__save.takeDirtyEntities()` + `__save.takeDirtyManifests()` and write every affected file, including every record still waiting from earlier frames. Idempotent / cheap to call when nothing is dirty.

writerFrame( ) → boolean

The continuous writer's frame: take the marks that landed, write one frame's budget of the oldest waiting records, and rewrite changed manifests. Answers whether records still wait, which has the engine run it again on the next frame whether or not another mark lands.

flushDirty( ) → void

Synchronous flush. Used by `wld.play()` immediately before the mode flip so every pending edit-mode mutation lands on disk before the VFS write-lock takes hold.

pendingRecords( ) → number

How many entity records the writer holds to write: the marks the engine has noticed, taken into the backlog, plus the records already waiting there. A caller about to wait for the writer reads this to know how long the wait is.

examples

if saver.pendingRecords() > 64 then log.info("the writer is behind") end

flushDirtyAcrossFrames( ) → void

Put every pending edit-mode mark into the dirty overlay the way the continuous writer does: one frame's budget of records at a time, with the frame handed back between shares, so the time the calling frame spends is set by the budget however many entities the marks name. Returns once the overlay holds every mark taken. Where the caller cannot yield the whole backlog is written in the one frame, as `M.flushDirty` does. The edit to play flip calls this before it takes the write lock.

examples

saver.flushDirtyAcrossFrames()

getInstalledWatcherId( ) → void

setInstalledWatcherId(id: ?) → void

argtypedescription
id?

startDirtyWriter( ) → void

Event-driven drain. Subscribes once to `__save.onMarksLanded(callback)`; the engine schedule fires the callback once per frame whenever the dirty set transitions empty to non-empty since the previous drain, and again on the frame after any run that answers records are still waiting. Zero per-frame cost on idle frames. A reactive subscriber on each transition (`writerFrame`) plus the synchronous `M.flushDirty` (which `wld.play()` calls just before the mode flip) covers: • single mark: recorded on the next frame • marks across multiple frames: taken on every frame one lands, each entity waiting once however often it is marked • more marks than one frame's budget: written over the frames after, oldest first • mode flip: `flushDirty` synchronously writes everything before the VFS lock

⌬ Types
PendingSceneEdit = {BlockedSceneEdit = { id: string, name: string?, reason: string }SceneEditDiscard = {
⌬ Metadata
descriptionLuau-side scene saver. Serializes the current scene state to scene.json v6 — authored intent only (no runtime player position, no captured camera transform). Wrapped by layers.active:save({to=name}).

Sub-parts

Everything contained inside this part. Assets are composite children (clickable cards). Files are leaf payloads. Expand any row to view its source.

Showing up to 3 levels deep. Open a part to see what it holds further down.

112+items
▣
module · born here
❒asset
# scene_saver.recordDiff The comparisons the saver's record diffs (`transformDiffLines`, `recordDiffLines`, the pending scene edits) are built from. - `diffEqual(a, b)` and `valuesEqual(a, b)` compare two record values, numbers within a tolerance that absorbs the f32 round trip of a live read. - `componentMatchName(t)` answers the name a component type matches by, whichever spelling the record used; `componentList(rec)`, `componentMatchNameOf(c)`, `componentInstanceLabel(c, typeName)` and `componentData(c)` read a record's components the way the comparison needs them, answering empty for a malformed record. - `selfWrittenFieldSet(entityId, componentType, instanceName)` answers the fields a component instance wrote about its own runtime. - `pushAttributeDiffLines(oldRec, newRec, push)` hands `push` one line per attribute added, changed or removed. ```luau local RecordDiff = require("@builtin::assetTypes.scene.shared.saver.recordDiff") assert(RecordDiff.diffEqual({ 1, 2 }, { 1.00001, 2 })) RecordDiff.pushAttributeDiffLines({ attributes = { hp = 1 } }, { attributes = {} }, print) -- - attribute.hp ```
▲ 0↑ born
▣
module · born here
❒asset
# scene_saver.serialize One live entity as the record a scene file carries, and the reads the saver asks of a live entity while it decides what to record. - `serializeEntity(id, parentById)` answers the record for a live entity: its name, parent reference, transform, components (each field value serialized), participation, network scope, render layer and attributes. A field holding a runtime GPU handle is reported once, with the remedy, and left out of the record. - `serializeFieldValue(v)` answers one component field value as a record states it, with the reason when the value has no record form. - `entityDisplayName(id)` answers the name an entity is shown by, and `entityIsTemporary(id)` whether it is marked as never saved. ```luau local Serialize = require("@builtin::assetTypes.scene.shared.saver.serialize") for _, e in ipairs(entity.findAll()) do print(Serialize.serializeEntity(e.id).name) end ```
▲ 0↑ born
·
assettype · born here
❒asset
# Module (asset type) A module is a reusable, stateless (or globally-stateful) Luau library that other code pulls in with `require()`. Modules are the unit of factored-out logic in a world: helpers, math libraries, format parsers, registries, shared algorithms. ## When to use one - You have logic you want to share across multiple components, services, scenes, or tools. - You have a registry, lookup table, or shared cache that callers all read from / write to. - You want a stable API surface (`M.fn`) other code can depend on. If the logic is bound to one entity, use a `.component`. If you want a named UI of fields rather than functions, use a `.preset`. ## Where it lives - Source: `/zero/source/.../<name>.module/` - Identity: the source path with `/` written as `.` and each segment's type suffix dropped — `/zero/source/demo/beats.module` is `demo.beats`, and `/zero/source/kit.package/beats.module` is `kit.beats`, the container's name kept and its `.package` suffix gone. `asset.create` returns this string, `asset.resolve(...).identity` reads it back, and the `require` forms below write it as `<identity>`. - Folder shape: - `init.luau` (or `init.lua`) — module body. **Required.** Returns a table (conventionally named `M`). - `README.md` — instance-level documentation. **Required.** - `.metadata` — agent-editable tags + free-form fields. **Required.** ## How to create one ```luau asset.create("module", "<name>") -- Creates: /zero/source/<name>.module/ -- init.luau (canonical `local M = {} … return M`) -- README.md (instance README template) -- `folder` places it in a subfolder of /source, and that subfolder -- leads the identity: `{ folder = "lib" }` produces `lib.<name>`. -- `into` authors it inside a resolved container (a package, a toolbox), -- whose own identity leads the same way: asset.create("module", "<name>", { folder = "lib" }) asset.create("module", "<name>", { into = asset.resolve("myPkg.package") }) ``` ## How it operates 1. **Registration.** Writing `init.luau` into a `.module/` folder indexes the asset and registers the module path with the resolver. 2. **Loading.** The first `require("<identity>")` runs the file's body and caches the returned table. Subsequent requires from anywhere return the **same table** — this is the shared-state lever for "module-local" registries. The body runs in an environment of the module's own: a name it declares without `local` is readable by the module's own functions and by no other script, so an export reaches its callers through the returned table and nothing else. Requiring a component's folder runs its chunk the same way, once, and its `update`, `start` and `public` stay in that environment; a component runs on an entity through `entity(id).component.add(name)`. 3. **Hot reload.** Editing the file re-runs the module body and folds what it returns into the table already loaded, so that table keeps its identity: a caller that captured it with `require` reads the new source's functions and values through the reference it already holds, and a later `require` returns that same table. A key the new source drops goes away; a key something set at runtime stays. The body's own `local`s start again from their initial values, so state that must outlive an edit lives in `modules.state()` (`guides { path: "core/scripting-and-tasks" }`). 4. **Identity resolution.** A module is reached by its `<identity>`, and where the caller lives decides which spelling of it resolves. For a module in a **world**, from a caller in that same world — another module, a component, a scene entrypoint, or an `execute` chunk: - `require("<identity>")` — the identity on its own, resolved against the caller's own root. It carries every folder segment, so a module created with `{ folder = "demo" }` is `require("demo.beats")`. - `require("@root::<identity>")` — the same root-relative resolution, written out. - `require(".sibling")` — a module in the same folder, by its bare name; `..name` steps up one folder, `...name` two. Logs, stack traces and hot-reload notices name that same module `@local.source.<identity>` — its key under the world source root. The `<identity>` tail of that key is the form above, so a trace reading `@local.source.demo.beats` is `require("demo.beats")`. For a module in a **library** — content under `/zero/source/libs/@<lib>/` — `@<lib>::` addresses that library's root from anywhere: `require("@builtin::modules.transform")`. Mounting a world as a library moves its files under such a root and flips the caller root to `@<lib>::`, which is what keeps the root-relative forms above resolving across the move. ## Discovery - `asset.list("module")` — every registered module. - `asset.inspect("<name>")` — public functions, source path, this type README. - `ls /zero/source/<name>.module` lists the files the folder holds. ## Authoring conventions - Return a single `M` table from `init.luau`. Top-level statements with side effects run on first require — useful for one-time initialization, dangerous if they touch the engine before it's ready. - Write exported functions as `typed function`, not `function`: the argument types are enforced at the call and the checker knows the signature, so a caller's mistake is reported where they made it. - Annotate exported functions with `--!desc` / `--!arg` / `--!return` / `--!example` so the LSP, `tools.list`, and `asset.inspect` surface them. A function with no `--!desc` is reachable only by someone already reading this file. - Declare a VALUE with `Field.<kind>(default, mode, description)`, and a table whose members are reached through a metatable with `--!members <TypeName>` — the latter gives the surface its accepted member set, its type, and its documentation from one table. See `guides { path: "core/authoring" }` for all four rules and `modules/api/engine/entity.module/members.module` for a worked member table. - Keep module state in module-local upvalues or in `modules.state()`. A name the body declares without `local` lives in the module's environment, which every run of the module shares: a function the previous run left in place reads what the reload assigned. Module-locals start again from their initial values with each hot reload; the table `modules.state()` returns is the same table on both sides of every reload. - Prefer focused modules over kitchen-sink modules. If two halves of a module have no shared state, split them. ## Common pitfalls - **Cyclic requires.** A `require` chain that loops will return the partial table (the half built before the cycle was detected). Design for one-way dependencies. - **Engine-time side effects.** Don't `entity.spawn` at module top-level — the engine may not be ready. Expose an `M.init` and call it from a scene entrypoint. - **Hot-reload + module-local cache.** A module that caches expensive computation in upvalues loses that cache on reload — fine for dev, but be aware in performance work. - **`init.lua` vs `init.luau`.** Either works, `.luau` is canonical. - **Writing one while play runs.** A `/source` write made while play is running lands on the play shadow, and a guarded play-exit discards it unless it is kept. One `.module` write shadows `init.luau`, the `README.md` and the `.metadata` together. `vfs.write(path, bytes, { durable = true })` skips the shadow, and `vfs.promotePlayShadow` keeps what is already on it — the `core/vfs` guide has the model and what each route costs. ## Related types - `.component` — for entity-bound state + lifecycle hooks. - `.service` — generates content you don't have yet (a mesh, sound, texture, …) via a metered provider. - `.tool` — for a single agent-callable function with a YAML schema. - `.package` — to group several modules + components + scenes into one shippable folder. ## In the Inspector Exports: what the module exports (name, kind, signature), parsed from its source without running it, with **Check** (the LSP's diagnostics for its entry file) and **Open**.
▲ 0↑ born
▣
module · born here
❒asset
# asset_create `asset.create` — the single, generic "instance a new asset of an existing type" API. The same entry point scripts and agents use; there is deliberately no tool wrapper, because agents author in Luau and call `asset.create(...)` directly. ```lua local inst = asset.create("material", "my_metal", { base_color = { 0.8, 0.7, 0.2 } }) -- → AssetRef: inst.path == "/zero/source/my_metal.material", inst.guid, plus the -- material type's ref methods. ``` `asset.create` makes a new *instance of an already-registered type* by running that type's `onCreate(name, opts)` behaviour hook (declared in its `behavior.luau`) and writing the produced files to `/zero/source/<name>.<typeName>/` — the one authored location, in every mode. While play runs, the play-mode write lock takes that write onto the **play shadow**: live in the session, listed by `vfs.playShadowPaths()`, and promoted or discarded on a guarded play-exit. The asset created in play therefore carries the same identity, path and `require` spelling it has in edit, and the session decides whether it stays. A create whose output the world **reproduces** on every load — one made from a component or a scene entrypoint — writes to the copy-on-write runtime store at `/zero/runtime/assets/<identity>.<typeName>/` instead, so the code that rebuilds it each load is its only source and the saved manifest never carries a second copy. ## Placement Four `opts` keys are consumed by the framework before the type's `onCreate` hook runs, and steer where the instance lands: | Key | Effect | |-----|--------| | `folder` | A relative subfolder under the source root: `/zero/source/<folder>/<name>.<typeName>/`. Groups a generator's output instead of accumulating it at the source root. | | `into` | A resolved container ref (`.toolbox` / `.package`) to author INSIDE — lands at `<container>/<name>.<typeName>/` and registers as a member. Edit-mode only. | | `dest` | An absolute destination path, owned by the caller (an importer building a `<name>.bundle/`). | | `overwrite` | Re-author an existing destination in place, keeping its `.meta` guid so every reference stays valid. | The `name` is always the asset's bare identity — the path goes in `folder`: ```lua local rock = asset.create("mesh", "rock", { positions = p, indices = i, folder = "terrain/props" }) -- → rock.path == "/zero/source/terrain/props/rock.mesh" ``` `dest` and `into` both take precedence over `folder`. `folder` decides the asset's identity, so it applies in every context: the runtime store is flat and spells that identity in one folder name. ```lua -- the same call from a component's awake() -- → rock.path == "/zero/runtime/assets/terrain.props.rock.mesh" -- → rock.identity == "terrain.props.rock", as it is from an execute ``` Types without an `onCreate` hook fall back to cloning their verbatim `template/` skeleton to the same destination, so `create` works for every type. Returns the created asset's `AssetRef` (`.path` / `.guid` plus the type's ref methods — the same interned instance `asset.resolve` returns) on success and raises (via `error`) on bad arguments or a failing `onCreate`. Installed onto the FFI `asset` namespace by the prelude (`M.installInto(asset)`). See `docs/specs/runtime-asset-copying.md`.
▲ 0↑ born
▣
module · born here
❒asset
# Field edits Scene authoring uses FieldEdits to keep structured field intent separate from generated data. Each edit carries a path, the value that field was based on, and the authored value, with explicit presence flags for additions and removals. ```lua local Fields = require("@builtin::modules.fieldEdits") local base = {position={0,0,0}, light={intensity=1}} local visible = {position={2,0,0}, light={intensity=1}} local edits = Fields.diff(base, visible) local rebuilt = Fields.compose({position={0,0,0}, light={intensity=3}}, edits) assert(not rebuilt.blocked) assert(rebuilt.value.position[1] == 2) assert(rebuilt.value.light.intensity == 3) ``` - `copy(value)` validates and copies structured data without sharing mutable tables. - `diff(before, after)` creates deterministic field edits. Objects merge by string key; arrays and asset references are atomic values. - `compose(generated, edits)` applies authored values and reports divergence from their stored baseline. Missing parents, missing edited targets and conflicting root additions block the result. Satisfied paths identify intent the generated value already matches. - `update(generated, previous, current, edits)` records a new authoring operation while retaining existing baselines and satisfied intent. An addition followed by its removal cancels that addition. - `updateChanges(generated, current, changes, edits)` records original observed changes against their merged visible result. A change already satisfied by the current generated value remains an authored intent for subsequent builds. - `keep(generated, edits, paths?)` acknowledges reviewed generated changes while retaining the authored values at the selected exact paths. Inputs are copied, edit paths are non-overlapping dense arrays of strings, and values are finite serializable data with bounded nesting. Scene-level hierarchy, reference and component validation belongs to the scene composer consuming these field results.
▲ 0↑ born
▣
module · born here
❒asset
# layers Top-level scene-management global + Scene proxy. § 17 step 9 of `docs/plans/2026-05-01-player-camera-unification.md`. Renamed from the legacy `scene.*` global to match `/runtime/layers/` and to give scene-management state (active root, additives, callback registries) a single anchor distinct from per-scene authoring. ## Global surface | Symbol | Type | Notes | |---|---|---| | `layers.active` | `Scene \| nil` | Current root non-additive scene as a proxy. Live read; updates on every load/unload. | | `layers.list()` | `{ Scene }` | Every loaded scene (root + additives) as proxies. | | `layers.find(ref)` | `Scene \| nil` | Lookup by `AssetRef<scene>` or layer-id string. | | `layers.is_loaded(ref)` | boolean | | | `layers.load(ref, opts?)` | — | Loads ref. `opts.additive = true` layers on top; default = non-additive (replaces root). | | `layers.unload(ref?)` | — | Unloads ref; nil = active root. | | `layers.reload(ref?)` | — | Equivalent to unload + load with same ref. | | `layers.onLoad(cb) -> handle` | function | Subscribe to any scene-load event. Returns handle for `offLoad`. | | `layers.offLoad(handle)` | function | | | `layers.onUnload(cb) -> handle` / `offUnload(h)` | function | Same shape for unload events. | | `layers.observe()` | `SceneObservation` | What every load did and what each loaded scene costs, in one read. | | `layers.lastLoad()` | `SceneLoadReport \| nil` | The last load's report; nil when nothing has loaded. | | `layers.lastUnload()` | `SceneUnloadReport \| nil` | The last unload's, under the name the layer was loaded with. | | `layers.loadHistory()` | `{ SceneLoadReport }` | The completed load reports still held, oldest first. | | `layers.problems(ref?)` | `{ SceneLoadFailure }` | What a layer failed to produce, and why. | | `layers.whyPartial(ref?)` | `(reason?, detail?)` | The nearest cause a layer is not whole, from a closed set. | | `layers.inventory()` | `{ SceneLayerInventory }` | What the engine attributes to each loaded layer. | | `layers.cost()` | `{ SceneLayerCost }` | Each layer's per-frame entrypoint tick, summed across the window. | | `layers.resetCostWindow()` | — | Open a new cost window. | ## Scene proxy surface A `Scene` proxy is a sealed table with fields and `:method` callables. | Field / method | Notes | |---|---| | `s.asset` | `AssetRef<scene>` | | `s.name`, `s.path` | informational (logging, debug) | | `s.guid` | the scene asset's guid — the layer's identity | | `s.sceneIdentity` | the identity of the scene asset the layer was loaded from (e.g. `@builtin::scenes.test_arena`), whichever VM loaded it | | `s.additive` | boolean — additive overlay vs root scene | | `s.parent`, `s.children` | additive ownership chain | | `s:root()` | walks `.parent` to the non-additive root; returns self for roots | | `s.players`, `s.camera` | populated by §17 step 10 (player + camera spawners) | | `s.entrypoint` | `AssetRef<script> \| nil` | | `s.visible` | boolean — read via field, toggle via `s:set_visible(b)` | | `s.state` | `"loading"` / `"loaded"` / `"ready"` / `"partial"` / `"unloading"` — the layer's lifecycle state. Both `"ready"` and `"partial"` are terminal: the load has settled. | | `s.ready`, `s.loaded` | booleans in lock-step with `state`. `ready` is true once the layer has settled, in either terminal state — it is the flag to wait on. | | `s.ok` | boolean — whether everything the scene declared was produced and the layer has run clean since. False whenever `state` is `"partial"`, including when the tick starts raising long after the load settled. | | `s.failures` | `{ SceneLoadFailure }` — every distinct failure the load and the layer's per-frame tick produced, each with the `count` of how often it repeated. | | `s:unload()` | unload this scene; additives use `layers.unload(layerId)`, roots use `layers.active:reload()` | | `s:reload()` | unload + reload same ref | | `s:set_visible(b)` | show/hide without unloading | | `s:load_additive(ref)` | load a sibling additive overlay; returns the new proxy | ## Example ```lua -- Switch root scene layers.load(asset.ref("@builtin::scenes.baseline_visual")) -- Subscribe to all loads local h = layers.onLoad(function(scene) print("loaded:", scene.name) end) -- Layer an HUD overlay on top layers.load(asset.ref("@builtin::scenes.hud"), { additive = true }) -- Iterate every loaded scene for _, scn in ipairs(layers.list()) do print(scn.name, scn.additive and "(additive)" or "(root)") end layers.unload() -- unload the active root layers.offLoad(h) ``` ## What a load did, and what a scene costs Every load writes one record at its own completion point; the readings below compose that record, so a caller that did not make the load can still ask what it did. A load that produced failures settles into `"partial"` rather than `"ready"`, and names the nearest cause from a closed set: `loaderRaised`, `entrypointCompileFailed`, `entrypointBodyRaised`, `entrypointRaised`, `buildRaised`, `entityFailed`, `parentMissing`, `parentRefused`, `parentAbandoned`, `componentUnresolved`, `componentRefused`, `subscriberRaised`, `updateRaised`. ```lua layers.load(asset.ref("scenes.arena", "scene")) -- Did it come up whole? if not layers.active.ok then local reason, detail = layers.whyPartial() print(reason, detail) for _, f in layers.problems() do -- `count` rises when the same failure happens again, so a per-frame -- tick that keeps raising stays one line. print(f.phase, f.reason, f.entity, f.component, f.hook, f.count) end end -- What did the swap do? local r = layers.lastLoad() print(r.name, r.outcome, r.durationMs) print("replaced", r.replaced and r.replaced.name) -- the root that went print("cascaded", #r.cascaded) -- the overlays with it print(r.entities.added, "arrived,", r.entities.removed, "left") print(r.phases.teardown, r.phases.instantiate, r.phases.settle) -- Which loaded scene is expensive? `totalMs` is a SUM across the window, so -- reset it first and read `avgMs` for the per-tick figure. layers.resetCostWindow() task.wait(1) for _, c in layers.cost() do print(c.name, c.avgMs, "ms/tick over", c.calls, "ticks") end -- What does each layer hold, and what does no layer claim? for _, l in layers.inventory() do print(l.name, l.entities, l.ok) end print(layers.observe().totals.unattributed) ``` The `scene` toolbox carries the same two readings as `scene.observe` and `scene.whyPartial`. ## Authoring surface `layers.*` is the sole public scene-management surface. It accepts AssetRefs (not paths), drives a callback registry, and constructs Scene proxies that carry players / camera / hooks. The previous `scene.*` global has been removed — its low-level FFI bindings now live behind `layers.*` and the Scene proxy methods.
▲ 0↑ born
▣
module · born here
❒asset
# Source operations Require `@builtin::modules.sourceOperations` to coordinate revision-checked changes to text files in a bound world. `capture(paths, directories?)` waits for local writes in the named scope and captures authoritative file revisions and directory membership. It includes containing asset inventories for their derived checksums. The returned snapshot is frozen and names its world and branch. `revisionToken(snapshot)` returns a deterministic token for the guarded world, branch, file revisions and directory membership. A later capture with the same guarded revisions has the same token; guarded changes invalidate it. It leaves the snapshot and source files unchanged. `source(snapshot, path)` reads a SourceEditing snapshot whose identity and storage checksum match the captured row. Use SourceEditing's call discovery and exact-range helpers to select the intended declaration and scope. ```lua local Ops = require("@builtin::modules.sourceOperations") local Source = require("@builtin::modules.sourceEditing") local path = "/zero/source/scenes/main.scene/build.luau" local snapshot = Ops.capture({path}) local source = Ops.source(snapshot, path) local calls = Source.calls(source, "asset.resolve") assert(calls[1], "select an asset.resolve declaration in this builder") local edit = Source.replace(calls[1].arguments[1], Source.literal("props.Table")) local plan = Ops.plan(snapshot, {{path=path, source=Source.preview(source, {edit})}}) -- Review plan.uploads and the selected declaration's affected outputs. local prepared = Ops.prepare(plan) assert(prepared.state == "prepared", prepared.reason) local result = Ops.commit(plan.operation, plan.scope) assert(result.state == "committed", result.reason) ``` `plan(snapshot, changes)` accepts disjoint text-file replacements and explicit `{path=..., remove=true}` removals. It preserves existing file identities, allocates identities for new files, derives candidate reference tables and recomputes containing asset checksums through the engine. Every mutation must be covered by the captured read set. Create asset containers through `asset.create`, then plan changes to their source files. Directory inventories participate in revision checks, including additions at previously absent paths. References are derived against the current asset registry and the captured prior reference table. `plan` enforces ordinary source-write permissions, including computed-output protection declared by an asset type. A compiler uses `materialize(snapshot, changes)` to include its computed outputs together with their inputs. Materialization keeps the same revision, directory, identity and reference checks and respects world and library write permissions. `prepare(plan)` first retains the exact request and its candidate pins, then uploads missing candidates under those pins. It returns prepared state once every candidate is available, already committed state after a completed operation, or failed state with the operation id retained. Repeated preparation uploads only missing candidates. Interrupted preparation is inspectable as awaitingCandidates, with the missing content hashes; resume with the retained plan or cancel it. `commit(id, scope?)` applies the retained request against its read set. A committed result acknowledges server persistence; scene activation observes the resulting files. `inspect(id, scope?)` returns committed, prepared, awaitingCandidates, cancelled or missing state. Prepared records include the exact request, and committed records include the receipt. Use the same operation identity after a lost reply. `pending()` enumerates this identity's unfinished operations in the bound world and branch. `cancel(id, scope?)` releases a prepared operation's candidates and makes its identity terminal. Failed commits retain their prepared record for inspection and explicit resolution.
▲ 0↑ born
▣
module · born here
❒asset
# entity_records Walks a live entity hierarchy into the flat record array the engine's entity templates use — `{ name, original_id, parent_id, position, rotation, scale, hidden, components }` per entity, root first, parents before children. `bundle` composes its `entity_template` through this module; scene builds compose their baked output through it. One record shape, one implementation. `compose(rootId)` walks one hierarchy. `composeMany(rootIds)` walks several, which is what a build needs — a builder may leave more than one unparented root. Temporary descendants are pruned; an explicitly-named temporary root is kept. `compose(rootId, nil, nil, true)` keeps the temporary descendants too, so the records hold every entity under the root.
▲ 0↑ born
▣
module · born here
❒asset
# presence Editor presence: who else is in this scene, where they are looking from, and what they are working on, for every client of the world in edit mode. People in their editors and agents in their sub-clients appear alike. Each client occupies one entity of its own, the way a player occupies its avatar in play: a temporary `PlayerOwned` root that the edit room relays like a player's subtree and the scene never records. It stands where the client's camera stands, and its `EditorPresence` component carries who the client is and what it is doing: its selection, the entities it is changing, and the call an agent is making. Every other client draws that participant on the entity it receives: a viewpoint marker in the participant's colour, a name label stating the activity, and an outline in the same colour around each entity the participant has selected or is changing. ```lua local EditorPresence = require("@builtin::modules.api.editor.presence") EditorPresence.ensureRegistered() -- the debug-viz provider that draws the others EditorPresence.tick(dt) -- publish this client's presence (the editor's per-frame tick) EditorPresence.noteChanged(entityId) -- this client is changing that entity now for _, p in EditorPresence.participants() do print(p.name, p.kind, p.activity) end local color = EditorPresence.colorOf("agent-1") -- the same colour on every client ``` | Call | Does | |------|------| | `ensureRegistered()` | Registers the provider that draws the other participants. Idempotent. | | `tick(dt)` | Occupies this client's presence entity, moves it to the viewpoint and states the selection, the entities being changed and the agents' activity. Runs from the editor's per-frame tick in edit mode. | | `noteChanged(id)` | Notes that this client changed the entity `id` now, so its presence states it is changing it for the next two seconds. | | `track(id, presence)` / `untrack(id)` | Starts and stops drawing the participant on the presence entity `id`. | | `participants()` | One row per remote participant this client draws: its name, kind and activity. | | `colorOf(name)` | The `{ r, g, b, a }` colour a participant named `name` is drawn in, the same on every client. |
▲ 0↑ born
▣
module · born here
❒asset
# Scene composition transactions `capture(folder, inputs)` reads a scene and its authored records under a world-bound source transaction. Capture build inputs before evaluating the builder, and pass the resulting document to `rebuild`. `edit` accepts saved entity records before and after an authored change. Both compose and validate before preparing an atomic update of the accepted generated base, intent records, and `scene.json`. `editChanges(snapshot, entities, settingsEdits, opts, retirePaths?, attribution?)` accepts compact observed field changes. Each entity row has `id`, `changes` from `FieldEdits.diff` over normalized records, and optional component `bookkeeping`. Settings use field edits over the scene fields. Changes retain their original observed values when retried against a newer snapshot. `retirePaths` names captured journal files removed atomically with the accepted composition; a newer journal revision makes the entire transaction stale. `editDocument` projects saved documents into this same field-change path. Journal consumers pass a session/sequence receipt to `editChanges` with the corresponding captured operation path in `retirePaths`. The coordinator checks the consecutive sequence, records its checksum in accepted state and retires that source atomically. Rebuilds preserve these receipts. `assetTypes.scene.shared.journal` handles operation creation, persistence, recovery and retry checks. `migrate` accepts a legacy visible document derived from the captured scene and overlay, together with a recovered generated base when available. It retains the canonical and visible documents as recovery history and records unresolved baseline identities. `incorporate` prepares source replacements together with selected intent retirement. Each selection names an entity (or scene settings), the captured intent revision, and exact field paths. The candidate generated value must reproduce every selected value. Unselected intents remain. Candidate source fingerprints, the generated base, build digest and artifact advance together. For a loaded scene’s `build.luau`, `scene:prepareSourceChange(sourcePlan, selections)` evaluates the candidate from `SourceEditing.prepareTarget`. It returns the coordinator plan together with the target’s scope and known outputs. Candidate evaluation uses isolated capture: persistent writes, external requests, deferred task work and foreign-entity changes are refused. Component lifecycle work on candidate-owned entities is captured and cleaned up. `awaitApplied(folder, revision, scope)` waits for an accepted complete revision to arrive through the local subscription. A committed result acknowledges backend persistence; this wait establishes local visibility. `commit` returns the durable source-operation result. Its operation identity supports inspection, retry and cancellation through `modules.sourceOperations`. Runtime scene activation observes the committed files. Structural conflicts return a plan with `ready=false` and preserve the last valid document.
▲ 0↑ born
▣
module · born here
❒asset
# resource_handle Recognises a **live GPU resource handle** — the value `renderer.mesh.create`, `renderer.material.create` and `renderer.texture.create` answer with. A handle is a table of the shape `{ kind = "<Category>Handle", category, guid, name }`, and its `guid` names a slot in the **running process's** GPU registry. It is minted when the resource is created and identifies that resource for as long as the registry lives. An `AssetRef`'s guid is content identity — it names a stored asset, and any session that can reach the world's content resolves it. The two look alike at a glance and answer differently at every **session boundary**, which is what this module exists to tell apart: - **The replication wire.** A peer runs its own GPU registry, so a handle's guid means nothing there. The sync layer drops such a value and the receiver falls back to the field's declared default. - **Durable content.** A scene record outlives the session that wrote it. The next session — a peer loading the world, or the same engine after a restart — mints its own registry, so a recorded handle names a resource that session never created. `asset.create(category, name, handle)` is the crossing: it stores the resource and hands back a ref whose guid does survive both boundaries. A record is also the medium a scene reloads through **inside** one session: a play flip unloads the live entities and rebuilds them from the same bytes a peer receives, and the handle is valid there. So a handle a record leaves out is **parked** in `modules.session` under the entity and component it came from — `park` as the record is written, `withParked` as it is applied. A load inside the minting session puts it back; every other session opens an empty store and the field takes its declared default. ## API ```luau local ResourceHandle = require("modules.resource_handle") local mesh = renderer.mesh.create(geometry) ResourceHandle.isLive(mesh) -- true ResourceHandle.isLive(asset.resolve("wall.texture")) -- false (an AssetRef) ResourceHandle.isLive("cube") -- false ResourceHandle.label(mesh) -- "mesh handle 'msh_grass_1'" -- Component field maps: `data` unchanged when it holds no handle. local data, dropped = ResourceHandle.withoutLive({ model = mesh, visible = true }) -- data -> { visible = true } -- dropped -> { model = "mesh handle 'msh_grass_1'" } -- What a record leaves out, held until no record names the entity. ResourceHandle.park(entityId, "@builtin::components.Model", { model = mesh }) local applyData, restored = ResourceHandle.withParked(data, entityId, "@builtin::components.Model") -- applyData -> { visible = true, model = mesh } -- restored -> 1 -- No record names the entity any more: nothing will put its handles back. ResourceHandle.forget(entityId) ``` `withoutLive` returns the **same table** when there is nothing to leave out, so the common case allocates nothing. When it does drop, the second return names each field and the handle it stood for, so the caller reports what happened at the point it happens rather than leaving the field to fail later. `park` replaces whatever that entity-and-component pair held before, so a field that stops holding a handle stops being restored with the ones that do. `withParked` fills only the fields the record has no value for — a record that carries a value is the authority on it. The scene saver calls `forget` for an entity once a save leaves no record naming it, so what is parked is the size of the entities the scene carries rather than of every entity a session wrote. ## Consumers `scene_saver` asks per field as it serialises an entity, reports the entity, component and field once per session, and parks the handles it left out. `scene_loader` asks per component as it applies a record: it reports the whole scene's count of recorded handles in one line — one generator run writes the same field on every entity it made, so a per-entity report would say the same thing hundreds of times — and puts the parked ones back. `dirty_hot_reload` applies a peer's body to a live entity and leaves the handles in it out.
▲ 0↑ born
◇
component · born here
❒asset
# Camera Manages viewport priority, render-to-texture, and capture. State is stored in the native `Camera` ECS component; the Rust camera system handles render scheduling and render targets. Public fields: `fov`, `near`, `far`, `priority`, `textureHandle` (the guid of the texture the camera renders into; empty = main viewport), `renderLayers` (which render layers this camera draws — a space-separated spec of names, e.g. `"all"`, `"all !ui"`, or `"default sky"`; `ui`/`sky`/`debug`/`EditorUI` are built-in layers), `postProcessing` (whether this camera runs the post-process chain), `debugChannel`, plus the behavior slot below. Methods: `:lookAt(target)` (entity id string OR `{x, y, z}` table), `:render()`, `:capture()`, `:setTargetTexture(tex?)` (a `renderer.texture.create` handle to render into, or nil for the viewport). ```luau entity(id).component.add("Camera", { fov = 90, priority = 10 }) entity(id).component.get("Camera"):capture() ``` The scene's play-mode camera is reached as `layers.active.camera`, a handle that reads and writes the fields above on whichever entity currently carries them. `layers.active.camera.entity` is the entity ref for that camera and `layers.active.camera.entityId` its id string, so a script that needs to attach something to the camera — an `AudioListener`, a child entity — goes through the ref: ```luau local cam = layers.active.camera cam.fov = 70 -- the camera's settings cam.entity.component.add("AudioListener") -- the entity carrying them ``` ## How the camera moves: `behavior` and `follow` A `Camera` does not move itself. `behavior` names a component that does, and setting it attaches that component to this entity. Clearing it detaches whatever was attached. ```luau local cam = entity(id).component.get("Camera") cam.behavior = asset.ref("@builtin::controller.orbital_follow", "component") cam.follow = playerBody ``` The attached behavior is an ordinary component on the camera's entity, so it is tuned the way any component is: read it with `component.get` and set its fields there. A `component.add` of the same type configures the instance the slot attached instead of adding a second one: the values it passes are written onto that instance and its proxy is returned. That add is the form a scene `build.luau` tunes the behavior with, because its values are saved with the scene and the loaded Camera's behavior takes them back. A key the component reads only as it is created (one of its `initKeys`, a private field) refuses the add and names the key. ```luau cam.behavior = asset.ref("@builtin::controller.third_person_follow", "component") local rig = entity(id).component.get("third_person_follow") rig.cameraDistance = 6 -- the same, as an add of the type the slot attached entity(id).component.add("third_person_follow", { cameraDistance = 6 }) ``` `follow` is the standard slot every shipped behavior reads. Set the follow target on the **Camera**, not on the behavior, so swapping behaviors keeps it. A behavior that finds its own `follow` field empty falls back to this one, which is what lets a rig keep tracking the player across a behavior swap. `followResolves` answers whether that slot names an entity that is live — `true` while it names a live one or names nothing at all, `false` once the target is despawned or the id names no entity. A rig whose target does not resolve holds its last pose, and this is the field that tells it from a rig posed correctly on a subject that has not moved. `camera.get` carries the same value beside `follow`, and a write naming an id with no entity behind it draws a warning where it lands. ```luau cam.followResolves -- false once the followed entity is gone tools.use("camera", "get", id).followResolves -- the same answer off the tool ``` The shipped behaviors live under `@builtin::controller.*`: `orbital_follow`, `third_person_follow`, `first_person`, `free`, `orbit`, `chase`, `isometric`, `rts`, `birds_eye`, `side_scroller`, `cinematic`, `menu`. ## Writing your own Any component can be a camera behavior. Write one that moves its own entity and attach it the same way: ```luau cam.behavior = asset.ref("MyChaseCam", "component") ``` `follow` lives on the Camera, so a behavior gets no property notification of its own when the target changes. Declare `onFollowChanged(newFollow, oldFollow)` to be told the moment it does — the Camera calls it on the component it attached, which is what lets a rig re-pose on the new subject at once. A behavior that reads `Camera.follow` on its own schedule declares nothing and is attached the same way. ```luau typed function public:onFollowChanged(newFollow: any, oldFollow: any) -- pose this entity against the new target end ``` To make it appear in the discovery catalog alongside the shipped ones, declare the `cameraBehavior` tag in the component's `.metadata`: ```json { "tags": ["cameraBehavior"] } ``` The tag governs **discovery**, not attachment. A tagged component is listed by `layers.active.camera.behaviors` and is what tooling offers when something asks "which camera behaviors exist"; an untagged component attaches just as well and simply stays out of that list. Tag the ones you want other people (and agents) to find. ```luau for name, ref in pairs(layers.active.camera.behaviors) do print(name, ref.identity) end ``` ## Texture colour space `textureColorSpace = "display"` applies the display transform when rendering into a texture. `"linear"` writes scene-linear values instead. `postProcessing` independently controls the effects chain in either mode. Floating-point targets retain values above one; normalized targets clamp to their representable range. The main viewport uses display encoding.
▲ 0↑ born
▣
module · born here
❒asset
# persist.origin Provenance / origin-context layer for the persist (create-flow) system. The problem persist must solve: when the agent freezes live play state into `scene.json`, it must capture **only** the things that would not otherwise be reproduced on the next load — i.e. creations whose stack root is the agent's ad-hoc `execute()` context. Anything created by `onLoad` / the scene loader or by a component callback re-runs on every load, so freezing it would duplicate it. This layer tags each live creation with its origin context (via wrappers installed over the entity API) and exposes the current origin, so the serializer can keep execute-origin entities and drop the rest. See `docs/specs/multiplayer/candidate-c-persist-flow.md`.
▲ 0↑ born
▣
module · born here
❒asset
# Authored scene copies Prepare a guarded copy of an authored scene with fresh contained-asset identities, relocated JSON references and exact static Luau references, preserved authored records, and updated source bindings. Preparation leaves the destination untouched. Unsupported or ambiguous internal reference forms are refused before commit. `prepare(source, destination)` returns a source transaction; `commit(plan)` persists it and observes the copied scene identity and artifact.
▲ 0↑ born
▣
module · born here
❒asset
# Scene record projection Require `@builtin::assetTypes.scene.shared.records`. `saved(record, componentGuid)` converts a persisted entity into `{value, bookkeeping}`. `captured(record, ids, componentGuid)` converts an entity-record capture and remaps captured entity references to persistent placement ids. The resolver receives a component type spelling and returns its stable identity; unresolved types must raise an error. Captured entities require an entry in `ids`. `value` holds editable fields with normalized transform and lifecycle defaults. Components are grouped by stable type identity: `named[name]` addresses one named instance, while `anonymous` preserves an ordered group. Saved `instance_index` markers are reconstruction metadata, and restoration emits them alongside named `instance_name` markers. FieldEdits treats that anonymous array atomically. Asset references compare by guid and type. Component references retain their entity, instance information and stable component type. `bookkeeping` holds build ownership attributes, component order, and captured reference paths and row-field declarations. These capture annotations describe reconstruction and stay outside field comparisons. Compose edits over `value`, then call `restore({value=composed, bookkeeping=current.bookkeeping}, componentName)` to produce the saved record. The reverse resolver supplies the current component type spelling. Surviving instances retain their order; added instances are appended in deterministic order. The module validates structured input before copying it. Duplicate named instances and missing identities fail before any scene mutation.
▲ 0↑ born
◇
component · born here
❒asset
# Player Identity tag for a player entity. Attaches the Rust `PlayerOwned` marker via `__native` in `awake` and removes it in `onDestroy`. No public state, no update loop, no visuals. ```luau entity(id).component.add("UserIdentity") entity(id).component.has("UserIdentity") -- true ``` Future rules (not yet enforced): - **Non-serialising.** Scene and world saves omit this component. Players are attached at runtime per peer, not inherited from the save file. - **Non-removable.** Once an entity is a player, removing this component is a no-op. A player stays a player for its lifetime outside an explicit identity swap. Visuals, cameras, input, and movement live in separate components (e.g. `PlayerController`, `PlayerTemplate` bundle content) so adding or removing them never affects identity.
▲ 0↑ born
▣
module · born here
❒asset
# scene_build Turns builder code into scene records, and scene records into scene entities. `run(builder)` opens an entity capture scope, calls `builder`, and composes every entity it created into records — then despawns them. The builder writes ordinary spawn code; what comes back is data. A builder that raises leaves nothing behind. What a component the builder attached creates while running its own lifecycle belongs to that component: the record names the COMPONENT, and the same lifecycle runs again wherever the record is put back. That is what makes a nested build one build — a placement the builder makes runs its own module and owns what it lands, and the build around it describes the placement. `reconcile(records, target, owner)` applies records to the scene under `target`, writing `owner` and the record's place in the hierarchy onto each entity it places as the `buildOwner` / `buildKey` attributes. A rebuild reads that pair back off the entities under `target` and lands on the same ones, so nothing has to be remembered between two rebuilds — the pair is in the scene, and a reload brings it back with the entity. It returns the ids it landed on, keyed by record identity. A record is identified by its place in the hierarchy — the chain of names from the build root down to it, `base/coin_3` — so dropping a line from the middle of a builder, inserting one, or emitting two entities in the other order leaves every other entity's id where it was. Children of one parent that share a name carry their rank among those, `base/crate#2`, counted in the order the builder created them. Together these are the bake: code in, durable static-id scene content out. ## Who states an entity An entity names the build that placed it, so `ownerOf(id)` answers which build owns it — and answers nil for an entity an author spawned. `sourceOf(owner)` names the file that build is written in, from the `source` the reconcile running it passed. That pair is what makes ownership legible where an author meets it. `notePreview(id, values)` records what an entity carries after an author's change and `takePreview(id)` reads it back exactly once, in the run that states the entity again, which is where what the author set and what the build says can be named side by side. ## What a rebuild keeps Every entity a build places records what that build stated about it. A rebuild whose code states a value the way the last build did leaves that value as the scene holds it, so a hand edit to a built entity (a move, a rename, an attribute, a component field) survives every plain rebuild. The scene's build answers for all of it: ```lua local report = layers.active.asset:build({ save = false }) print(report.message) -- opens with one line counting all four lists ``` `kept` lists the hand edits to entities the build placed, `driven` the fields another component writes at runtime, `handAdded` the components and attributes added by hand to built entities, and `handPlaced` the entities no build placed. `layers.active.asset:build({ restate = true })` lands every value `build.luau` states over the kept ones. `reconcile(records, target, owner, { restate = true })` is the same at this level, and its third return value is the `kept` list. `kept` covers every value a build states about an entity it placed: its name, internal flag, transform, active flag, lifecycle mode, replication scope, synced flag, attributes and components. A restate writes exactly those: a number the scene holds within the precision a saved record keeps, and the bookkeeping a component writes about its own runtime, stay as they are. `drift()` reads the entity's own values alone; the build's report is the whole answer. `guides { path: "core/scenes-as-code" }`, "What a rebuild touches", gives the shape of every entry.
▲ 0↑ born
▣
module · born here
❒asset
# persist.serializer Live-state serializer driver for persist — the play-mode half of the scene-save story. `scene_saver.serializeEntity` already turns a live entity into the v6 per-entity body, but it is only ever fed ids from the edit-mode dirty-mark queue (empty in play). This driver supplies the play-mode side: it enumerates the active layer's live entities, filters them by **provenance** (execute-origin only, via `persist.origin`) and **authored-ness** (`scene_saver`'s spawner-managed / temporary filter), and assembles a loadable v6 scene body. Lighting stays entity-primary: it comes from `Light` components on the live entities, not from a separate lighting block.
▲ 0↑ born
▣
module · born here
❒asset
# profiler Module Public Luau surface over the `__profiler` Internal FFI namespace — frame-level profiler capture, EMA stats, and named blocks. ## Purpose Wrap the raw `__profiler.*` FFI namespace in a typed Luau table that gets auto-injected as `_G.profiler` via the prelude. Captures persist to VFS at `/zero/runtime/profiler/<label>.json`; EMA stats and named blocks land back in the engine's `SystemProfiler` resource so they're visible in captures and in the in-engine profiler panel. ## Usage ```luau -- Capture a window of frames. One recording runs at a time engine-wide, -- and several agent sessions share one engine, so a refused start is -- answered with who holds it rather than with the bare fact that -- something does. if not profiler.startCapture("hitch") then print("held by", profiler.recordingHold()) return end task.wait(2) local json = profiler.stopCapture() print(json) -- Read live EMA stats. `avg_frame_ms` is the block's share of the -- current frame; `avg_ms` is what one run of it costs. for _, s in ipairs(profiler.stats("schedule.*")) do print(s.name, s.avg_frame_ms, s.avg_ms, s.max_ms) end -- Named blocks profiler.begin("ai.tick") runTick() profiler.finish("ai.tick") -- Or scope it with measure() (handles errors) local result = profiler.measure("ai.tick", function() return runTick() end) ``` ## Exports - `profiler.startCapture(label?) -> boolean`. Takes the one engine-wide recording slot for the calling caller. False when another recording holds it; `profiler.recordingHold()` then names what does. - `profiler.recordingHold() -> string?`. JSON `{ label, owner, started_at, held_seconds }` for the live recording, or nil when none runs. `owner` is the caller key that took the slot, the same value `vfs.currentAuthor().id` reads for the calling caller, so a caller refused the slot tells its own earlier recording from a sibling session's. - `profiler.stopCapture() -> string?`. JSON capture summary. `callers` carries each caller's share of the window's queued Luau, and each frame carries its own split as `agent_by_caller`. - `profiler.isCapturing() -> boolean` - `profiler.lastCapture() -> string?` - `profiler.stats(pattern?) -> { ProfilerStat }` — live EMA stats. Each entry carries two averages: `avg_ms` averages one RUN of the block, folded when the block runs, so a block that has stopped running keeps the last value it saw; `avg_frame_ms` averages one FRAME, folded every frame including the frames the block did not run in, so it is the block's share of the current frame and falls back to zero once the block stops running. - `profiler.counters(pattern?) -> { ProfilerCount }` — unitless per-frame tallies kept apart from the millisecond timings of `stats`, for example the engine thread's waits on the process-wide lock under `system.schedule.frame.process_local_waits`, and the queue submissions the renderer and the UI made under `system.schedule.frame.queue_submissions` (each one is a fixed cost to the GPU driver, whatever it carries), and the uploads the renderer's device staged: the buffer and texture writes under `system.schedule.frame.queue_buffer_writes` and `system.schedule.frame.queue_texture_writes`, and the staging buffers the device created to carry them under `system.schedule.frame.staging_buffer_creations` (the device stages a frame's writes in chunks it reuses, so a frame whose writes fit the chunks it holds creates none; in a browser's WebGPU the browser stages the writes and the three read zero), and the maps of the renderer's statistic readbacks under `system.schedule.frame.gpu_readbacks`, with each readback's own maps under that key followed by its name: `hiz.occlusion_counts` and `instanced_compaction.submitted` (the counts `renderer.cullStats().occlusionTested` / `occlusionCulled` and `renderer.drawStats().compactedDrawn` answer, read back on the two rendered frames after each read and one frame in 60 otherwise), `cluster_cull.counters`, `visibility.resolve_counters` and `gpu_timing.readback` (so `system.schedule.frame.gpu_readbacks.hiz.occlusion_counts` counts the occlusion counts' maps), and the profiler names resolved by hashing them under `system.profiler.name_lookups`: a measurement's name is resolved the first time it is recorded and found by its key after that, so a frame recording only measurements it has recorded before counts none (a `profiler.begin` / `profiler.measure` block finds its key by the Luau string naming it, so a name a script builds again after its last string was collected is resolved again). `frame` is the tally of the frame in progress, `prev_frame` the last complete frame's, `max_frame` the highest complete-frame tally and `total` the tally since the profiler started. - `profiler.begin(name)` / `profiler.finish(name?) -> number?` — paired named blocks, LIFO by default. - `profiler.measure(name, fn) -> ...` — scoped block around a function.
▲ 0↑ born
▣
module · born here
❒asset
# Observed scene operations SceneJournal stores field edits with their original observed and proposed values. Each entry belongs to a world, branch, scene GUID and unique editing session. Session sequence numbers order operations under `scene_dirty/operations/<session>/<sequence>.json`. `session()` creates a session token. `create(snapshot, session, sequence, previous, current, resolvers, attribution?)` normalizes the two scene documents and records their field differences. `append(sceneFolder, entry)` writes an immutable entry; it returns the path and whether an accepted receipt already settles the entry. Reusing a pending sequence with different bytes is refused. The latest retired sequence can be retried with identical bytes; older retired sequences are refused. `createChanges(observed, session, sequence, entities, settingsEdits, attribution?)` records incremental normalized changes. `observed` carries the scene GUID, world, branch and accepted revision. The live scene writer uses this form for changed entities and settings. `visible(sceneFolder, resolvers, session?, previous?)` composes pending entries over the locally accepted artifact. An optional session restricts which pending values are shown. Conflicts identify their journal operation. Structural conflicts retain the supplied previous complete document, or the accepted artifact. The read leaves all source files unchanged. `resolveFields(snapshot, path, resolutions, resolvers)` chooses values for exact changed fields. Each selection has an entity id (omit it for scene settings), a field path and a `recorded` or `accepted` choice. Other proposed changes remain. Every remaining conflict must be resolved before the transaction is ready, and duplicate or absent field selections are refused. `review(snapshot, resolvers)` returns captured operation rows, their recorded changes, current conflicts, and whether an earlier entry in the same session must settle first. `resolve(snapshot, path, choice, resolvers)` prepares a resolution of one whole reviewed operation. Choose `recorded` to retain its proposed fields against the reviewed accepted values, or `accepted` to keep those accepted values. Structural validation still applies. The resulting transaction checks the reviewed source revisions and retires only that operation with its receipt; a stale resolution leaves it intact. `prepare(latestSnapshot, path, resolvers)` recovers the captured entry, verifies its identities and predecessor, and prepares a scene transaction. Commit it with `SceneCoordinator.commit`, then use `awaitApplied` before reading the accepted revision locally. Successful commit updates the artifact and authored intents, removes the consumed journal entry, and records its sequence and checksum atomically. A stale operation, validation failure or field conflict retains the entry for inspection and resolution. The accepted scene state retains one receipt per editing session.
▲ 0↑ born
▣
module · born here
❒asset
# json JSON encode/decode library for Luau. Encodes Lua values to JSON strings and decodes JSON strings back to Lua values. Used for communication with the Rust side of the engine, the VFS read/write bridge, and any wire-format that needs JSON. Pure Luau, no engine dependencies. Compact and pretty-printed encoders, plus a hand-rolled decoder that streams the input by position so it works under WASM as well as native. ## Exports - `Json.encode(value: any, indent?: string, currentIndent?: string) -> string` — compact encode. Functions / unknown types and NaN/Inf encode as `null`. - `Json.encodePretty(value: any, indentStr?: string) -> string` — pretty-printed encode with sorted object keys (diff-friendly). - `Json.encodeArgs(...: any) -> string` — encode varargs as a JSON array. - `Json.decode(str: string) -> any` — decode a JSON string. Returns the decoded value, or `nil` + error message on failure. ## Usage ```luau local Json = require("@builtin::modules.json") local widget = { type = "button", text = "Click Me" } local compact = Json.encode(widget) -- '{"text":"Click Me","type":"button"}' local pretty = Json.encodePretty(widget, " ") local decoded = Json.decode(compact) local v, err = Json.decode("oops") -- v = nil, err = error message ``` ## Notes - Object keys are sorted alphabetically in both encoders for consistent output across runs. - Numeric keys on objects are stringified at encode time (JSON has no numeric keys). Pure-integer key sets get detected as arrays via `isArray` and encoded with brackets. - NaN, +Inf, -Inf encode as `null` — JSON has no representation. Round trips through `decode` recover `null` (Lua `nil`), so they don't preserve. - Unicode `\uXXXX` escapes decode to UTF-8 by hand to stay WASM-safe. Only the BMP is covered; supplementary planes via surrogate pairs are not. - Functions encode as `null`. - Decode is character-streamed — no regex, no `string.match` patterns on the whole input — so the line-and-column information needs to be reconstructed from the position offset.
▲ 0↑ born
▣
module · born here
❒asset
# frame_budget The share of a frame that cooperative engine work spends before it hands the frame back. Some engine work is one long sequence of small units run on a task: a scene load applies one record after another, and the world-load `onRegister` sweep fires one instance's hook after another. Run back to back in one frame, a long sequence holds that frame for its whole length, and nothing else the engine does (answering a call, drawing, input) happens until it is through. Asking `due()` before each unit and yielding when it answers true runs the same work across as many frames as it needs, a slice of each. ```luau local FrameBudget = require("@builtin::modules.frame_budget") for _, item in ipairs(items) do FrameBudget.yield() process(item) end ``` - **One slice per frame, shared.** Each caller is charged the time it runs between its readings of the slice in a frame (`task.runTime`), whoever runs it, and the charges of every caller add up to the frame's slice. What the rest of the frame runs (systems, components, other tasks) is not charged, so a busy frame still gives the slice to the work. Two long pieces of work running at once share a frame's slice rather than taking one each. - **A unit is never split.** The first unit of a frame always runs, so a frame holds at most the slice plus the one unit that crossed it. A unit that is itself long stays one piece of work. - **Code that cannot yield runs in its frame.** `due()` answers false outside a coroutine that can yield, so a caller called from such code does all its units in the frame it was called in, as it would without the budget. A caller that does work of its own around the yield asks `due()` and calls `wait()`, which yields to the next frame and opens that frame's slice as the caller resumes; `yield()` is the two together. Work whose unit is a native call that bounds itself by time is told how long it may run: `remainingMs()` answers what is left of the current frame's slice, or nil for code that cannot yield, which then runs the whole of its work in the frame it was called in. The publish gate's script check runs this way: ```luau local block, checking = __spacetime_vcs.scriptPublishBlock(FrameBudget.remainingMs()) while checking do FrameBudget.wait() block, checking = __spacetime_vcs.scriptPublishBlock(FrameBudget.remainingMs()) end ``` ## The background lane Some work is started early so it is ready when asked for, and no picture waits on it, such as the search text an editor reads ahead of the first `tools.search`. `background(fn)` runs `fn` on the background lane: every `yield`, `due` and `wait` its work makes on that coroutine hands the frame back for as long as a hold on the lane stands, and takes its slice as usual once none does. ```luau FrameBudget.background(function() for _, path in ipairs(paths) do FrameBudget.yield() warm(path) end end) ``` Work the picture waits on holds the lane while it runs: `holdBackground(key)` until `releaseBackground(key)`. A world's entry holds it under `"world entry"` from the moment the world is ready (its `onRegister` sweep) until its `onWorldLoad` has returned, so the frames of the entry carry the world's load and the background work runs in the frames before and after it. A hold no one releases lapses after `BACKGROUND_HOLD_SECONDS` (or the seconds it was taken for), with a warning naming its key, so work on the lane always runs in the end. `backgroundHolds()` lists the keys holding the lane now, and `backgroundWaits()` answers whether the calling code's next unit waits for it (a walk with a clock of its own asks it between its pieces). `FrameBudget.DEFAULT_MS` is the slice a call that names none spends. `spent()` answers how much of the current frame's slice is gone. A caller with a budget of its own reads `mark()` where its work starts. `ranMs(mark)` answers how long it has run since, none of the time it spent suspended; `cpuMs(mark)` answers the CPU time of its work since: those stretches and the stretches of every coroutine it resumes, and every one they resume, while it waits on them, which a machine whose cores are taken by other work does not stretch, so a bound that refuses work for running too long refuses the same work on a loaded machine as on an idle one, and work handed to a coroutine as work run in place. A task it starts and waits for runs while it is suspended, which `cpuMs` does not count.
▲ 0↑ born
▣
module · born here
❒asset
# vfs_async_read Pure-Luau wrapper that grafts a yielding `vfs.read` over the engine's sync `vfs.read` binding. The `vfs` API module (`modules/api/engine/vfs`) calls `M.installInto(vfs)` on itself as it loads, so every VM holding `vfs` (user VMs and the trusted VM alike) reads through it; user code never requires this module directly: it just calls `vfs.read(path)` and the wrapper handles cache misses by yielding the running coroutine until the bytes resolve. ## Exports - `M.installInto(vfs: VfsNamespace)`: replace `vfs.read` (and `vfs.readBounded` and `vfs.list`) with the yielding wrapper. No-op when the target lacks `read` and `readAsync` as functions. - `M.CANNOT_WAIT`: `"VFS_READ_CANNOT_WAIT"`, the code a read raises when it names a known file whose bytes have not arrived and the reading code cannot yield. Types: - `VfsNamespace = { read?, readAsync?, readBounded?, list? }`: the engine's `vfs` global shape. ## Usage ```luau -- The `vfs` API module installs the wrapper onto itself as it loads: require("@builtin::modules.vfs_async_read").installInto(vfs) -- After install, every `vfs.read` call yields on cache miss, and a read -- that finds no bytes answers nil and the reason: local bytes, why = vfs.read("/zero/source/some.asset") ``` ## Where a read can wait - On a miss the wrapper schedules `vfs.readAsync(path, opts)`, which fires `BlobProvider::fetch` under the hood, and `task.await`s the promise. A wait that resolves without bytes answers nil and the reason: the sentence the engine's read gave beside its nil, or, for bytes that were on their way, the one it gives once the wait ends. - Waiting takes a yield, so it happens where `coroutine.isyieldable()` is true: a task, the component hooks the engine runs on a coroutine (`awake`, `start`, `update`, `fixedUpdate`, `onEnable`, `onOwnerChanged`, `onSyncReceived`), a replicated event's handlers, and a world or scene entrypoint's hooks. - Where it is false (`onDisable`, `onDestroy`, `onPropertyChanged`, a received synced-function call, a `vfs.watch` callback, a module's top level while `require` runs it, a metamethod, a `table.sort` comparator, a VM's main thread) a miss the engine's read answers `nil, "pending"` for (a file `vfs.exists` knows that is no directory and whose bytes can still arrive, a render surface being captured, a `/runtime/sync` node whose current reading the next frame assembles) raises `VFS_READ_CANNOT_WAIT: vfs.read("<path>") ...`, naming the path and the places it can be read from. Every other miss answers nil and the reason there as everywhere else. - `vfs.list` answers a directory whose current listing the next frame assembles (a `/runtime/sync` directory nothing has read lately) after waiting that frame where the code can yield, and raises `VFS_READ_CANNOT_WAIT: vfs.list("<path>") ...` where it cannot. - `installInto` is idempotent in the trivial sense: calling it twice re-wraps `vfs.read` around the already-wrapped function, which is fine but pointless.
▲ 0↑ born
▣
module · born here
❒asset
# Scene composition Require `@builtin::assetTypes.scene.shared.composition`. `compose(generated, intents, settingsEdits, options, previous?)` computes a candidate scene document. Each entry in `intents` is keyed by persistent entity id and contains `edits` over the value returned by `SceneRecords.saved`. An entry can also carry captured `bookkeeping` for component additions. `settingsEdits` addresses the scene document with its entities field removed. Options supply `componentGuid(typeName)` and `componentName(guid)` resolvers. `externalEntities` explicitly identifies entities supplied by an enclosing composition. The result contains `candidate`, `document`, `blocked`, `conflicts`, and satisfied paths per entity. Field conflicts retain authored values. Missing edited entities, invalid parent graphs, dangling declared entity references and missing component-reference targets block the candidate. A blocked result returns an independent copy of `previous` as its visible document. Its edits remain the caller's durable input. `planMigration(canonical, visible, generated?, options)` converts a legacy saved scene and its composed dirty state into a generated baseline, entity intents and settings intents. It validates that composing these inputs preserves every normalized visible entity and scene setting. Supply a recovered build baseline where available. Without one, retained content becomes authored additions and the plan reports unresolved baseline identities. Legacy deletions remain explicit intents even when only the saved record supplies their prior values. Build identity metadata survives on retained additions. The function writes no files; the migration coordinator owns source snapshots, revision checks, recovery and activation. The composer validates entity references declared by captured `entity_fields` annotations and component-reference envelopes. Component schema and asset-specific validation belong to the caller's semantic validation pass. The function computes data; the caller chooses when to commit and activate that data.
▲ 0↑ born
·
assettype · born here
❒asset
# assetType (asset type) An asset type says what a kind of asset is: the files an asset of the kind is made of, the code that gives it behaviour, the interface a person uses to work with it, and how it is found, pictured and checked. A `<typename>.assetType/` folder declares a type called `<typename>`, and every `<name>.<typename>/` folder is an asset of it. `assetType` is itself a type, so a type is an asset like any other: it is created, versioned, installed and contributed the same way. ## Making a type `asset.create("assetType", "<typename>")` scaffolds a type. The type registers the moment its folder is written, anywhere under `/zero/source/`, and from then on `<name>.<typename>/` folders are its assets. An asset of a type that is not registered yet is not an asset, so make the type first. A type folder is a root: `~.x` in any file inside it names the type's own `x`, wherever the type is installed. Keep the type's whole implementation inside its folder and reach it that way. What a type folder must hold is set by this type's own `type.yaml`, the same way every type's `type.yaml` sets what its assets hold. `asset.validate` on a type reports anything missing, and the scaffold produces a folder that passes. ## What sets your type's schema (`type.yaml`) `type.yaml` is the structural contract every asset of the type is checked against. The keys below are the ones the engine reads; a key not listed here is ignored. | Key | What it controls | |---|---| | `suffix` | The folder suffix of the type's assets. Defaults to `.<typename>`. | | `primary` | The file or files that are the asset's content: what the asset resolves to when it is used. | | `entry_files`, `primary_aliases` | Further files that load as the asset or share its identity. | | `is_container` | Whether the suffix stays in the identities of what the asset holds. | | `is_opaque_walker` | Whether a walk of the tree stops at the asset and registers only its primary. | | `plural_dir`, `classify_exts` | Loose files in a folder of this name, or with these extensions, are assets of the type. | | `is_text_default` | Whether the asset's files are text unless stated otherwise. | | `broadcasts_subassets` | Whether assets nested inside an asset of the type are found by their own short name. | | `root` | `true` makes every asset of the type a root, so `~.x` inside one names that asset's own `x`. | | `contract_ref: { field, category }` | A key of the primary file that names another asset the asset is built on. | | `files: { required, optional }` | The files an asset holds (below). | | `allow_unlisted` | `false` reports every file the type does not declare. | | `aliases_from: [{ file, key }]` | Further names an asset answers to, read from its own files. | | `settings:` | Typed per-asset settings (below). | | `inspector:` | The interface people use for the type and its assets (below). | | `preview:` | How an asset is pictured (below). | | `references:` | Names of other assets the type's files carry (below). | | `indexing:` | What search reads for each asset (below). | | `describe:` | How a missing description is generated (below). | ### The files an asset holds Each entry under `files.required` or `files.optional` is `{ path, kind: file | folder, one_of_group?, content? }`. `*` in a path matches within one segment. Entries that share a `one_of_group` are satisfied by any one of them. `content` lists checks on the file's bytes: | Check | Passes when | |---|---| | `non_empty` | the file has content | | `json_field { key, type?, non_empty?, accept? }` | the JSON holds the key (a dotted path) with that shape | | `yaml_field { key, type?, non_empty?, accept? }` | the same for YAML | | `described { keys?, headings?, message? }` | the file says what the asset is (see "Ask your assets to say what they are") | Each check takes a `message`, which is what the author is told when it fails. ### Settings `settings: { values, schema: { type | instance, export? } }` gives every asset typed settings. `values` names the file an asset's values live in; `schema.type` declares one schema for the whole type, and `schema.instance` names a file each asset declares its own in. `asset.create` writes the resolved values, and a write to the values file reaches `onChange` with what changed. ## The human side (`inspector.luau`) A type is used by people through its Inspector, and without one the type is usable only from code. The Inspector is the type's interface for a person: for a procedural graph it is the graph editor, for a kit it is the tracking view of what the kit holds and what is still missing. Writing it is part of writing the type. `inspector: { type, instance? }` in `type.yaml` names the files. `type` is the interface for the type itself and for its assets; `instance` names a file an asset may ship to add its own. Each returns: ```luau return { sections = { { id = "graph", title = "Graph", build = function(ctx) ... end } }, replace = { contents = { ... } }, -- replace a default section by id order = { "graph", "contents" }, -- section order } ``` `build(ctx)` receives the asset and builds the section. A section's `title`, and the text its rows carry, is a string or an `i18n.text` value read from a catalog (`i18n.text("inspectors.mesh.title")`), which follows the editor's language. `asset.create("assetType", ...)` scaffolds a working `inspector.luau` to start from. `asset.validate` on a type that names no inspector, or names one that does not load, reports it. ## Say what your type is for (`about.yaml`) Every type carries an `about.yaml` that says what the type is for, in three plain-language texts: ```yaml intent: "Sets the typeface of the text in your game: UI, labels, signs and 3D lettering." function: "A font, from a TTF, OTF or WOFF file, made available to all of Zero's text, including 3D text." usage: "You choose it by name for your UI or your text, and the text appears in that typeface." ``` - **`intent`** is what the type is for in someone's game, in the words a game maker uses. Up to 800 bytes. - **`function`** is how it works today: what an asset of it is made of and what the engine does with it. Up to 1500 bytes. - **`usage`** is what a person or an agent does with one. Up to 1200 bytes. Write about the type as a whole, not one asset of it, and say what it does now. Agents read these texts to learn when to reach for the type and how; people searching find the type by them. The scaffold starts each text with `[[describe]]` and a note on what to write; a text that still starts that way counts as not written. `services describe` can write them. They show up in `asset.inspect` as `detail.about`, in `assetTypeRef:getAbout()`, at the end of the type's `types/<typename>` guide, and at the top of the type's Inspector. ## Ask your assets to say what they are A type can require each asset to carry a description: a file, written by the author or generated, saying what the asset is and what it is for. Declare it with a `described` check: ```yaml files: required: - path: "README.md" kind: "file" content: - { check: "described", message: "Say what this widget is for and where it goes." } ``` A file that is missing, empty or still has a line starting with `[[describe]]` fails the check, and so does a YAML file whose named `keys` are missing or over their `max_bytes`, or a Markdown file missing a named heading. `asset.validate` names the asset, and a push that carries it is refused with `message`. `services describe` finds such assets and writes their descriptions. `describe:` steers what is generated: ```yaml describe: guidance: "Say what the widget shows, what input it reacts to, and where it fits in a UI." examples: - "<identity of an asset of this type whose description reads well>" ``` An example is an asset of your type whose description is a true, written one, never a placeholder and never the asset being described; with none, leave `examples` empty. A type whose assets the model cannot perceive (a sound, a motion) says so in `guidance`, and the description is drawn only from what the asset's own files state. Start the file unwritten in your `template/`: a heading and one line starting with `[[describe]]`, never example content, which every new asset would carry verbatim. A type whose assets say what they are through their own content (a mesh, a texture) declares no check. ## The type's code (`behavior.luau`) `behavior.luau` returns one table. These are the keys the engine reads from it; anything else on the table is the type's own. ### What the type's assets can do | Key | Shape | What the engine does with it | |---|---|---| | `ref` | `{ [name] = function(self, ...) }` | Every function becomes a method on every asset of the type: `ref:name(...)`. | | `modules` | `{ [name] = require(".x") }` | Code the type's assets reach with `asset.containing(__FILE__).modules.<name>`. | | `events` | an events schema | Events an asset fires. Readers subscribe through `ref.events.<name>:connect(...)`; the type fires them. | | `refShapes` | `{ [method] = function(self) -> (typeExpr, source?) }` | The result type of a `ref` method for one particular asset (below). | | `namePattern` | a Lua pattern | The names `asset.create` accepts for the type's assets. Default `^[A-Za-z][A-Za-z0-9_]*$`. | | `derive` | `{ output, module, compute, fingerprint, matches?, required?, timeout? }` | A file computed from the asset when it is committed; a `required` one holds the commit until it lands. | Some `ref` names mean the same for every type, and the engine calls them: - **`instantiate(self, target?, opts?) -> (root, idMap)`**: puts the asset in a scene. Defining it is what makes the type scene-instantiable. Its return is a contract (below). - **`inspect(self) -> detail`**: what `asset.inspect` reports for the asset beyond the common record (below). - **`preview(self, opts)`**: renders the asset's picture; it wins over the type's `preview:` recipe. - **`previewMaterial(self) -> { material }`**: the material a `surface` preview paints. - **`saveDefinition(self)`**: writes the asset's in-memory edits back to its files in edit mode. Some names are answered for every asset before the type is asked, so a `ref` entry of the same name is never reached: `getSource`, `getBytes`, `getText`, `exists`, `deps`, `canInstantiate`, `canPreview`, `preview` (which calls yours when you define one), `meta`, `runtime`, `events`, `typeRef`, `modules`, and the residency flags `has_backing_asset`, `has_runtime_changes`, `cpu_resident`, `gpu_resident`. ### When the engine calls the type | Hook | Signature | When it runs | |---|---|---| | `onCreate` | `(name, opts) -> { [path] = contents }` | On `asset.create`. Returns the files of the new asset, laid over the type's `template/`. Annotate `opts`: its type is the schema `asset.create` checks the caller's arguments against. A `""` key writes a single loose file. | | `onReplace` | `(dest, files) -> files` | On an `asset.create` over an asset already at `dest`, after `onCreate`, so what the author added can be carried into the replacement. | | `onRegister` | `(self)` | Once per asset, when it first registers: when the world loads, on a live `asset.create`, and when an asset arrives later (installed, pulled, synced). This is what makes an asset take effect the moment it is written. | | `onChange` | `(ref, change)` | On every write to a file inside an asset of the type. `change` is `{ path, asset, type, kind, origin }`, plus `settings` and `settingsChanged` for a write to the settings values file; `kind` is `edited` or `seeded`, `origin` is `local` or `remote`. Filter on `change.path`. Writes to one asset arrive in order, each after the hook for the one before has finished, and the hook may wait on what it reads. A write the hook makes inside its own asset arrives later like any other, so keep the hook convergent. | | `onDependencyChange` | `(ref, change)` | When a module the asset's code requires is re-run, or a name the asset carries starts resolving. `change.kind` is `dependency`. | | `onDelete` | `(ref)` | When an asset of the type leaves the world: deleted, or the world it is in is swapped away. | | `validate` | `(ref) -> { { code, message, severity?, path? } }` | On `asset.validate`, after the `type.yaml` checks, for the type's own rules. `severity` defaults to `error`. Runs on every push, so a type's rules hold at publish. A hook that fails reports `validate.hook_failed`. | ### Settings hooks | Hook | When it runs | |---|---| | `settingsDefaults(name, opts, asked) -> table` | Once per `asset.create`, for values the schema's defaults do not give. | | `settingsCorrections(values, name, opts) -> (corrections?, reasons?)` | On every settings write. A correction to a field the writer named refuses the write, giving the reason. | | `settingsDerivedDefaults(values, explicit, changed)` | On a settings write, for fields derived from others. | | `normalizeSettings(values, stored?)` | On every read and write, to map an earlier spelling of a field to the current one. | | `settingsHealed(values, stored)` | On a read, to repair stored values. | | `settingsSchemaProblem(schema, ref) -> string?` | When an asset's own schema is read; a returned string refuses the schema. | `creationInputs` (`{ [opt] = true | selectorOpt }`) names `asset.create` options that are inputs to the type rather than settings. ### Reloading An edit to `behavior.luau` reaches existing assets with no restart. ## How an asset is pictured (`preview:`) | `kind` | The picture is | The type provides | |---|---|---| | `instantiate` | the asset placed in the preview studio through its `instantiate` | `ref.instantiate` | | `image` | a file the asset holds, fitted to the frame | `source: { files }`, the candidate files in order | | `surface` | the asset painted on a mesh | `mesh`, and `ref.previewMaterial` | | `custom` | whatever `ref.preview` returns | `ref.preview` | ```yaml preview: kind: surface mesh: sphere ``` `angle: [yaw, pitch]` turns the camera. The studio owns the light, the backdrop and the framing, so an asset pictures the same in every world. A type with no `preview:` that defines `instantiate` still answers `ref:preview()`. A picture is rendered when something asks to view it (the editor showing the asset, `ref:preview()`), when previews are generated on request (`assets.generatePreviews`), and when the asset is staged. An edit alone renders nothing. The picture records what it was made from, so a stale one is told from a current one. ## The names your type's assets hold (`references:`) Where your type's own format or API carries the name of another asset, declare it here: ```yaml references: keys: # a top-level key of the type's files that holds a name, or a table of them - { key: "shader", type: "shader" } args: # an argument of a call, counting from 1 - { call: "g:node", position: 2, type: "procNode" } fields: # a field of the options table a call takes - { call: "fx.beam", field: "material", type: "material" } ``` `type` is the category the name resolves in. `asset.declareReferenceKey`, `asset.declareReferenceArg` and `asset.declareReferenceField` declare the same from a script. ## What search reads (`indexing:`) `indexing:` is how assets are found. Search reaches an asset only through what its type's `indexing:` names, and adds nothing of its own. That includes you: an asset indexed badly is one you cannot find later, when you are making something else and need it. A type whose assets are never looked for says `indexing: none`. ```yaml indexing: data: # text, read from the asset's files - role: identity extractor: verbatim source: { file: "README.md" } - role: interface extractor: symbols source: { files: ["init.luau"] } - role: implementation extractor: verbatim source: { files: ["init.luau"] } content: # the picture - role: identity modality: image source: { file: "preview.png" } facets: # what a query filters and re-ranks by file: ".metadata" rigged: { type: bool, from: { components: { contains: SkinnedModel } } } roughness: { type: number, filter: range, from: { file: "mat.yaml", field: "floats.roughness" } } ``` **Roles.** Each `data` and `content` entry has a `role`, and a query matches in two slots: - **`identity`**: what an asset is. For anything you look at, the picture; for anything else, the text that says what it is. - **`capability`**: what it does and how it is made. `interface` (the signatures or shape a caller uses) and `implementation` (the body) both count as capability, and a type with code usually declares both. Ask what a person would recognise an asset of the type by, and what it does, and point each role at the file that says it. For text, that is the asset's description: the file your template starts with a `[[describe]]` block, which the author writes or `services describe` fills in (see "Ask your assets to say what they are"). **`data`.** `extractor: verbatim` embeds a file's text as it is; `symbols` embeds the declared signatures of code. `source` is `{ file }` or `{ files: [...] }`. Never index a file name (`source: { field: name }`): one or two words carry nothing to match. **`content`.** `modality: image` embeds the picture itself, so a text query matches what the asset looks like. A type whose assets are looked at makes the picture its identity. **`facets`.** What a query narrows and re-ranks the matches by. `file` is embedded whole, so keys nobody declared in it are still found. Each further key is computed for every asset and states a value for every one, including `false`: a file can imply "not rigged" but never say it, and a query separates rigged from unrigged only because the unrigged ones say `rigged: false`. A key is `{ type: bool | number | enum, filter?: range, from }`, where `from` is `{ file, field }` (a field of one of the asset's files, dotted for nesting), `{ metadata: key }`, or `{ components: { contains | any_outside } }` (what the asset's entities carry). ## What blocks a push A push checks every asset it carries against its type. These refuse it: - a required file that is missing, a `content` check that fails, a `described` file not written yet; - an error from the type's `validate`. Unlisted files and warnings do not. A `described` file blocks only a push that changes the asset. When a push is refused, the message names each asset and what it lacks. `world.publishBlockers()` and `zm status` read the same list without pushing. ## Typing a result from the asset (`refShapes`) A `ref` method whose result depends on the asset (one entry per child folder, per row of a config, per control a map declares) can only be annotated as `{ [string]: Thing }`, which accepts any key. `refShapes` states the result for each asset instead: a function returning a Luau type expression for that asset, and the module whose types the expression uses. Read the asset; never run it. ```lua -- inputMap.assetType/behavior.luau M.refShapes = { activate = function(self): (string, string) local names = {} for _, control in ipairs(M.ref.controls(self)) do table.insert(names, control.name .. ": Handle") end if #names == 0 then return "", "" end return "{ " .. table.concat(names, ", ") .. " }", "@builtin::modules.zinput.scheme" end, } ``` With it, `map.jump:onPressed(fn)` is checked and `map.noexisting` is reported. A per-asset type applies where the call site names which asset: a component field whose default names it, or an annotation `AssetRef<"inputMap", "<identity>">`. `asset.resolve(name, category)` carries only the category, and a name computed at runtime carries nothing, so calls on those keep the declared return. The type belongs to the call's result: index it there, or bind what you need from it there, and the names stay checked. `require("@builtin::assetTypes.assetType.shared.refShapes").published()` shows what was published: `categories` (each type's method surface) and `returns` (per-asset results by identity, then method). An identity missing from `returns` was never published. Results are recomputed when an asset is written; at most 64 assets of one type are read. ## Putting an asset in a scene (`ref.instantiate`) A caller writes one piece of code against every instantiable type, so what `instantiate` returns is fixed: ```lua local Instantiable = require("@builtin::assetTypes.assetType.shared.instantiable") M.ref = { instantiate = function(self, target, opts) local root = Instantiable.root(self, target, opts) -- compose the asset under `root`, now return Instantiable.result(self, root, idMap) end, } ``` **In.** `target` is the entity the asset lands under, or onto for a hierarchy type; with none, the type spawns a new root. `position`, `rotation` (three numbers in degrees, or a quaternion), `scale`, `name` and `temporary` mean the same for every type: `Instantiable.root` applies them to a root you make, `Instantiable.place` to one you adopt. Document any options of your own in the hook's `--!arg` docs. **Out.** `(root, idMap)`. `root` is the composed root, live when you return: build synchronously, never defer to a component's `awake`. `idMap` maps each original id to the spawned one, so a component that composes the asset again on every load can pass it back in and keep references into the composition. Every call goes through `Instantiable.result`, which refuses a root that is not live and turns a missing `idMap` into `{}`. ## Inspecting an asset (`ref.inspect`) `asset.inspect(ref)` builds a record every asset shares (`identity`, `name`, `guid`, `source`, `typeName`, `typeDefinitionPath`, `scope`, `origin`, `description`, `tags`), then calls `ref.inspect(self)` for the type's own `detail`: ```lua M.ref = { inspect = function(self) -- read the asset's own files, never runtime state: inspect works on an asset -- that has not been compiled, uploaded or registered return { --[[ what makes this asset what it is ]] } end, } ``` Without it, the record is the shared part alone, and an agent inspecting the asset learns nothing of what it is. A hook that raises leaves `detail` empty and sets `record.warning`. ## Discovery - `asset.list("assetType")`: every registered type. - `asset.inspect("<typename>")`: the type, its README and its `about.yaml` texts. - `asset.resolve("<typename>", "assetType")`: the type itself.
▲ 0↑ born
▣
module · born here
❒asset
# validate Pure validator for `asset.create` opts — checks a creation `opts` table against the schema produced by `asset.createSchema(typeName)`. Schema in, validated opts (or a teaching error string) out. Consumed by `asset.create`, which calls `M.validate` before forwarding opts to the type's `onCreate`. No FFI, no VFS. Errors are teaching errors: they state what was wrong at which path (`opts.cfg.size`), what was expected, and then render the full creation-parameter contract so the caller can fix the call without reading the type's `behavior.luau`. Unknown parameter names get an edit-distance "did you mean" suggestion. ## Types - `ValidatorShape` — `{ kind, literals?, item?, members?, fields?, indexer? }`, the recursive shape encoding. - `ValidatorParam` — `{ name, shape?, optional?, default?, desc? }`. - `ValidatorSchema` — `{ kind, desc?, example?, open?, error?, params?, indexer? }`. ## Exports - `M.validate(schema, opts?) -> (validatedOpts?, err?)` — validate and default-fill an `opts` table against an `asset.createSchema` result. Returns `(validatedOpts, nil)` on success (a new table for `schema`-kind schemas) or `(nil, err)` where `err` is a bare teaching error string. The caller owns presentation. Handles the `legacy`, `error`, `none`, and `schema` schema kinds. - `M.shapeName(shape) -> string` — render a single parameter shape (literal unions as `"png" | "jpg"`, arrays as `{ T }`, tables as `table`, primitives as their kind, nil/malformed as `any`). Used by `asset.describe`. - `M.renderContract(schema) -> string` — render a schema's creation-parameter contract as one aligned line per parameter, with an `[extra keys]` line for open schemas. ## Usage ```luau local V = require("@builtin::assetTypes.assetType.shared.create.validate") local out, err = V.validate(schema, { bytes = png }) if err then error("asset.create: " .. err) end ```
▲ 0↑ born
▣
module · born here
❒asset
# assetType runtime The machinery that reads a `<typename>.assetType/behavior.luau` and turns its declarations into live behaviour. A `behavior.luau` is a table of declarations — `ref` methods, `modules` shared code, an `events` schema, a `namePattern`, `refShapes`, and the `onCreate` / `onRegister` / `onChange` / `onDelete` / `validate` / `settingsDefaults` / `settingsCorrections` hooks. None of it runs itself. This is the code that runs it, and it lives here because reading a type's declarations is the **assetType type's own behaviour** — the same way `material.assetType` owns what a `.material` does. ## Sub-modules | Module | Reads | Responsibility | |---|---|---| | `ref.module` | `ref`, `modules` | Builds every `AssetRef` envelope, attaches the shared metatable, and dispatches per-type methods and shared modules onto it. Also owns edit-mode write-behind persistence. | | `create.module` | `onCreate`, `namePattern`, `validate` | Backs `asset.create`: resolves the type, checks the name against its pattern, runs `onCreate` (or clones `template/`), and writes the instance. `validate.module` beside it checks a creation `opts` table against the type's `createSchema`. Before running `onCreate`, resolves the type's settings chain (below) and hands the result to the hook as `opts.settings`. | | `settings.module` | `settings`, `settingsDefaults`, `settingsCorrections` | A type's declared settings schema and where its values live (`type.yaml`'s `settings:` block). `Settings.resolveForCreate` layers the schema's own defaults, the type's own `settingsDefaults(name, opts, asked)` hook (a DEFAULT the schema can't express — a name-derived convention, a fact read off a payload the caller handed it, a value following a field `asked` shows a higher layer naming; any later layer overrides it), the Preset Manager, an importer's settings, the caller's `settings` table, then the type's own `settingsCorrections(values, name, opts)` hook (a CROSS-FIELD CONSTRAINT the schema can't express, applied last over the fully merged values). A correction over a default/preset/importer value lands silently; a correction that would override what the caller's own `settings` table explicitly asked for refuses the creation instead, naming the field and why. Neither hook is ever handed a table to mutate; each returns what it wants changed (`settingsCorrections` may also return a `{ [field]: reason }` table for that refusal message). `Settings.cloneValues` deep-copies a values table — `create.module` hands `onCreate` a copy, never the table it stamps. `Settings.valuesOf` / `writeValues` read and write an asset's own values file. | | `typeYaml.module` | `type.yaml` | The folder a type lives in and its parsed `type.yaml`, decoded once per content version of the file. `settings.module` and `inspector.module` read their blocks through it. `about` / `aboutAt` read a type's `about.yaml` (`intent`, `function`, `usage`), leaving out any key that is empty or still starts with `[[describe]]`; `aboutRows` and `aboutMarkdown` lay the texts out for the Inspector and the `types/<name>` guide. | | `inspector.module` | `type.yaml`'s `inspector:` block | Which inspector a type declares for its assets (`inspector: { type: ... }` or `{ instance: ... }`), and the declaration that file returns — the sections it adds, the generic ones it replaces, and their order — or the problem that kept it from loading. | | `onRegister.module` | `onRegister` | Fires an instance's once-only registration hook — on arrival, on world load, and for `@builtin` library instances. | | `changeDispatch.module` | `onChange`, `onDelete`, `refShapes` | Routes a VFS source write or a folder delete to the enclosing typed asset's type hooks. | | `refShapes.module` | `refShapes`, `ref` | Publishes what every `AssetRef<category>` answers to, so a member read on an asset-typed value is checked against that category's real surface. | | `events.module` | `events` | The per-asset event runtime behind `ref.events`. | | `instantiable.module` | `ref.instantiate` | The instantiation contract an asset type opts into: `root` stands the root entity, `place` applies the base placement opts to one the type adopted, and `result` shapes the `(root, idMap)` every `instantiate` returns. Also registers the `sceneInstantiable` field-constraint validator. | | `declaration.module` | `type.yaml` | The type's decoded declaration, read once and held until a write inside an `.assetType` folder; what every question about a type as a whole reads. | | `previewRecipe.module` | `type.yaml` `preview:` | How an instance's `preview.png` is produced: which of the four recipes (`instantiate`, `image`, `surface`, `custom`) the type declares, and what that recipe needs. | ## Reaching it ```luau local rt = require("@builtin::assetTypes.assetType.shared") rt.ref.loadTypeBehavior("material") ``` Fields resolve lazily. The prelude requires the sub-modules directly, in dependency order (`ref`, then `changeDispatch`, `onRegister.install()`, `refShapes.install()`) — a table that required all six on load would take that ordering away from it.
▲ 0↑ born
▣
module · born here
❒asset
# scene runtime What a `.scene` folder does when it is loaded, saved, built, or swapped. A `<name>.scene/` carries `scene.json` (the entity arrangement), an optional `build.luau` (the code author's half), an optional `entrypoint.luau`, and a `scene_dirty/` overlay of unsaved edits. Reading, composing and writing all four is the scene type's own behaviour, so it lives in the type. ## Sub-modules | Module | Responsibility | |---|---| | `loader.module` | Reads scene.json v6 and v7, refuses versions it does not know, composes canonical + dirty overlay, spawns the entities, and runs the entrypoint. | | `saver.module` | Persists scene state as a delta overlay against the canonical file, promotes that overlay on save, and answers which edits are still pending. | | `build.module` | Runs a `build.luau` inside an entity capture scope and composes what it spawned into scene records. | | `observe.module` | The read side — the per-scene counts and costs `layers` and `loader` write. | | `swapOrchestrator.module` | Drives multiplayer room transitions and the scene-swap gate, so a local swap does not leak despawns to peers in the old room. | ## Reaching it ```luau local rt = require("@builtin::assetTypes.scene.shared") rt.saver.hasPendingSceneEdits(layers.active.name) ``` Fields resolve lazily. The prelude installs `swapOrchestrator` and starts `saver`'s dirty writer in a fixed order, which a table that required all five on load would take away from it.
▲ 0↑ born
·
assettype · born here
❒asset
# Component (asset type) A component is a reusable behavior + data unit attached to an entity. The component file declares public fields, methods, and lifecycle callbacks; the engine instantiates one per attachment and drives its hooks every frame. Components are the primary unit of gameplay code — movement, AI, animations, UI panels, network sync, and per-entity state all live here. ## When to use one - You need per-entity state that survives across frames. - You need per-entity behavior driven by engine lifecycle hooks (`start`, `update`, `onDestroy`). - You want to expose configurable fields to designers / `entity.spawn` callers / scenes without code changes. If you need a stateless pure-function library, use a `.module`. If you need to generate content declaratively, use a `.service`; compute work belongs in a `.computeShader`. ## Where it lives - Source: `/zero/source/.../<Name>.component/` - Identity: `<Name>` (the `.component` suffix strips from the identity, but the folder retains the suffix on disk). - Folder shape: - `init.luau` (or `init.lua`) — declares the component. **Required.** - `README.md` — instance-level documentation. **Required.** - `.metadata` — agent-editable tags + free-form fields. **Required.** The structural contract is enforced by `asset.validate` against this type's `type.yaml`. ## How to create one Always start with the scaffold — it guarantees the folder shape matches `type.yaml` from the moment it's on disk: ```luau asset.create("component", "<Name>") -- Creates: /zero/source/<Name>.component/ -- init.luau (canonical declare + public block + hooks stubbed) -- README.md (instance README template) ``` After scaffolding, edit `init.luau` to fill in the actual fields and behaviour. The component registers as soon as the file is written — hot-reload is automatic; no engine restart required. ## How it operates 1. **Registration.** Writing `init.luau` into a `.component/` folder indexes the asset, mints a guid (in the `.meta` sidecar), and surfaces the type to the engine's component registry. Same-frame visible to `entity.component.add`. 2. **Attachment.** `entity(id).component.add(name, data)` (or the batch form `tools.use("entityOps", "addComponents", targets, { [name] = data })`) attaches an instance. The `data` table's fields override the component's declared defaults; omitted fields take the declared default. The `data` values are copied into `public.*` BEFORE `awake()` runs. 3. **Lifecycle.** The hooks run in this order: - `awake()` — same frame as attach, immediately after the data copy. Initialise state, set up native bridges, register with other systems. Other components on the same entity may not have awoken yet. - `start()` — the next frame. Every component on the entity has completed `awake()` by then, so cross-component setup is safe here. - `update(dt)` — same frame as `start()`, then every frame after, for as long as the gameplay clock is running. `fixedUpdate(dt)` runs once per physics step, straight after it, with the step's `dt`, so a frame can run it several times or not at all; `editorUpdate(dt)` runs while the clock is held. Which of them ticks is decided by `engine.mode` AND `engine.paused` together — see **Per-frame hooks and the two axes** below, and `man modes`. - `onDestroy()` — when the component is removed via `component.remove` or when the owning entity is despawned. Release resources, drop native ECS components. For the full list (`onEnable`, `onDisable`, `fixedUpdate`, `editorUpdate`, `onPropertyChanged`, `onAssetReload`, `onModuleReload`, `onBeforeSave`, …) and the timing details, **read `man components` — that is the canonical lifecycle reference. Generic component patterns (declare blocks, public fields, sync, ECS bindings, typed methods) all live there.** 4. **Public surface.** Fields declared in `public` are inspectable, editable in the inspector, and persist with the entity. Methods are declared on that same table with a colon — `typed function public:method(args)` — and every one of them takes the component ref as `self`, so a caller invokes it with a colon too: `entity(id).component.get(name):method(args)`. That is the one calling convention a component ref answers to: the reflection methods `ref:fields()` / `ref:getField(name)` / `ref:setField(name, value)` take the same spelling. A function the chunk declares at file scope (`function helper()`) is the component's own, reachable from its hooks under its bare name; `public` is what the ref carries. 5. **Settings and the Inspector card.** `type.yaml` names the component's settings schema as `settings: { schema: { instance: "init.luau", export: "public" } }`: the `public` table IS the schema, and its values live on each entity the component is on, so `asset.schema(componentRef)` answers the Field descriptors and no values file exists. On an entity the Inspector shows the component as a card whose generic `settings` section edits those fields as typed controls, each edit checked against the schema and written to the live component as one undoable operation. `type.yaml` also names `inspector: { instance: "inspector.luau" }`: a component that ships an `inspector.luau` beside its `init.luau` shapes its card — the file returns `{ sections, replace, order }` (see `assetTypes/README.md`, "`inspector:`"), each section's `build(ctx)` reading `ctx.component` (the live component) and `ctx.selection.id` (the entity). `SceneModule.component/inspector.luau` replaces `settings` with the bound module's declared inputs. 6. **Events.** An `events` block declares typed moments the component fires from inside — `events = { onHit = Event({ dmg = Field.number(0, NoSync) }) }`. The owner fires with `events.onHit:fire({ dmg = 10 })`; outside code subscribes through the proxy's subscribe-only facade, `entity(id).component.get(name).events.onHit:connect(fn)` (`:once` / `:wait` too). Pass `Sync` as the event's second argument to replicate a fire to other peers. Events let one component react to another's moments without polling. Full reference: **`man components`**, Events section. 7. **Hot reload.** Editing `init.luau` reloads the component definition; live instances are reconciled with the new declaration in-place, and each crosses one lifecycle boundary: `onDestroy()` ends the running life, then `awake()` and `start()` construct the one that replaces it, and `update()` resumes on the next frame. `public` and `private` values carry across; module-level `local`s belong to the chunk, so the new life starts with them at their initial values and its `awake`/`start` are what fill them in. Whatever the old life handed elsewhere — a subscription, a shared registration, a spawned entity — is released in `onDestroy`, the call that pairs with the construction about to run. An edit that leaves the source byte-identical crosses no boundary at all. A field carries its value across when something stated it: a key of the `data` table the instance was attached with (a scene record's data is that table), or a write made since. Every other field takes the edited source's declared default, so changing a default reaches every instance that left the field to it. An instance that states its own value for a field whose default the edit changed keeps that value, and the reload logs a `[HotReload]` warning naming the field, the entity, the value kept and both defaults. ## Discovery - `asset.list("component")` — every registered component (engine + world + libraries). - `asset.inspect("<name>")` — identity, scope, public fields, lifecycle methods, source path, and this type README. `detail` carries the component's declared `fields`, `methods`, and `events`, parsed from its source — `tools.use("assets","describe", "<name>")` renders it as markdown, so a component's events are visible without opening its source. - `ls /zero/source/<Name>.component` lists the files the folder holds. - `entity(id).component.list()` — components attached to an entity. ## Per-frame hooks and the two axes `engine.mode` (`"edit"` / `"play"`) and `engine.paused` (the gameplay clock) are independent, and the per-frame hooks read them separately: - `update(dt)` / `fixedUpdate(dt)` tick while **the gameplay clock is running** — whichever mode the engine is in. - `editorUpdate(dt)` ticks while **the clock is held** — because the mode is `edit`, or because play is paused — and also alongside `update` whenever edit is running. Which gives four states, and this table is the contract: | `engine.mode` | `engine.paused` | hooks that tick | |---|---|---| | `edit` | `true` | `editorUpdate` | | `edit` | `false` | `update`, `fixedUpdate`, `editorUpdate` | | `play` | `false` | `update`, `fixedUpdate` | | `play` | `true` | `editorUpdate` | Entering edit sets `paused = true` and entering play sets it back to `false`, so a session that never writes `engine.paused` only ever sees rows 1 and 3 — which is where the shorthand "`update` is the play tick, `editorUpdate` is the edit tick" comes from. It is the two default rows of a four-row table, not the rule. **Row 2 is the one that surprises.** With `engine.paused = false` in edit — set by hand, or by a tool that advances the gameplay clock to sample it, such as a `capture.collage` given a `duration` — `update` and `editorUpdate` both fire, once each, every frame. A component that declares both and routes them to one body therefore runs that body **twice per frame**: ```luau local function drive(dt) ... end function update(dt) drive(dt) end function editorUpdate(dt) drive(dt) end -- twice per frame in row 2 ``` Nothing reports the doubling; it reads as an animation running at twice its intended speed. To get exactly one tick per frame in every row, ask which axis you meant and answer it in the body: ```luau function update(dt) drive(dt) end function editorUpdate(dt) if engine.paused then drive(dt) end end ``` A scene `entrypoint.luau` declaring the same two hook names is dispatched by the scene loader, not by this table: there `update` needs play AND running, `editorUpdate` needs edit, and play-paused ticks neither. Read `man scenes` before carrying a component's rule to an entrypoint. ## Authoring conventions - Annotate public fields with `--!desc` so the inspector + LSP show documentation alongside the field name. - Declare lifecycle hooks (`start`, `update`, `onDestroy`) explicitly rather than relying on engine defaults — readers can see at a glance what runs when. - Keep `update(dt)` cheap — it runs every frame per instance. Move heavy work to `start`, to a `.computeShader`, or behind a job (`man jobs`). - For multi-entity batching, prefer `tools.use("entityOps", "addComponents", targets, components)` over a loop of `entity(id).component.add` — it takes many targets and many components in one call. ## Common pitfalls - **Naming.** The component identity is the folder stem with `.component` stripped (e.g. `<Spinner>.component/` → identity `Spinner`). Don't repeat the suffix in the identity passed to `component.add`. - **`init.lua` vs `init.luau`.** Either is accepted (it's a `one_of_group: "entrypoint"`), but `.luau` is canonical. Mixing both in one folder is invalid. - **`update` in edit mode.** It runs whenever the gameplay clock is running, and entering edit stops that clock — so `update` is idle in a freshly-entered edit session and ticks again the moment anything sets `engine.paused = false`. Use `tools.use("wld", "play")` to test update logic under the mode it ships in; `tools.use("wld", "edit")` to return. See **Per-frame hooks and the two axes** above. - **Renaming.** Renaming the folder changes the identity. References to the component are recorded against the component itself, so renaming or moving it does not break them. ## Related types - `.module` — pure-function library, no per-entity state. - `.service`: a declarative content generator, no per-entity state. - `.preset` — a captured component configuration, replayable into `component.add` without re-typing fields. - `.bundle` — entity hierarchies that may carry component attachments as part of the bundle. ## In the Inspector Fields: its execution order, hooks, and how many methods and events it declares, then the fields its `public` table declares as read-only typed controls holding their defaults, with **Check** and **Open** for its entry file. Instances: a link to every entity carrying it. What its card shows on an entity comes from the component's own `inspector.luau`.
▲ 0↑ born
▣
module · born here
❒asset
# persist The create-flow persist namespace. Persist freezes live play state into a scene the way the agent's ad-hoc `execute()` context created it — its north star is "what you see is what you get": it preserves the exact live state, capturing only what would **not** otherwise be reproduced on the next load (so re-runnable `onLoad` / loader / component output is never duplicated). `require("modules.persist")` re-exports the three subsystems; each can also be required directly. - [`origin`](origin.module/README.md) — provenance / origin-context layer: which live creations trace to the agent's `execute()` stack (so only those are frozen). - [`serializer`](serializer.module/README.md) — enumerates the active layer's live entities, filters by provenance and authored-ness, and assembles a loadable v6 scene body (the play-mode half of scene-save). - [`player_camera`](player_camera.module/README.md) — freezes the live player avatar and camera into the scene's player/camera config.
▲ 0↑ born
▣
module · born here
❒asset
# ref Entity proxy metatable layer. Replaces the per-call closure-pushing entity proxy (`push_entity_proxy` in `crates/zero_scripting/src/ffi_callbacks.rs`) with a thin Luau metatable that calls into the `__native_transform.*` / `__native_entity.*` flat FFI bindings. The user-facing API is unchanged: `entity(id).localPosition = { x, y, z }`, `entity(id).despawn()`, `entity(id).name`, etc. all still work. After warm-up, steady-state allocation is **zero**: same `id` returns the same proxy (weak-valued cache), sub-proxies are lazily allocated once and cached on the proxy, and each method binding is a closure created on first access and cached as `__m_<name>`. ## Exports - `M.build(id: string) -> any` — entity proxy factory. Returns a cached proxy when one exists; otherwise allocates and caches a fresh one. Also bound as `_G.__build_entity_proxy` for the Rust-side `push_entity_proxy` call. - `M.EntityProxyMT`, `M.PositionProxyMT`, `M.RotationProxyMT`, `M.LocalPositionProxyMT`, `M.LocalRotationProxyMT`, `M.LocalScaleProxyMT`, `M.LossyScaleProxyMT`, `M.TransformProxyMT`, `M.AttributeProxyMT` — exported metatables; consumers that want to extend the proxies can read them, but the proxies are sealed externally (every `__newindex` errors). ## Usage ```luau local EntityProxy = require("@builtin::modules.api.engine.entity.ref") local proxy = EntityProxy.build("some-entity-id") proxy.localPosition = { 0, 1, 2 } proxy.localRotation.eulerAngles = { 0, 90, 0 } print(proxy.name) proxy.despawn() ``` ## Notes - **World vs local space** (post Bevy hierarchy migration): - `.position` / `.rotation` are world-space — engine back-computes `Transform.{translation, rotation}` on set, decomposes the world matrix on get. - `.localPosition` / `.localRotation` / `.localScale` are parent-relative direct reads/writes on `Transform.*`. - `.lossyScale` is read-only world-space scale (approximate under non-uniform parent scale + rotation; mirrors Unity). - `.worldMatrix()` answers the whole composed object-to-world matrix as 16 column-major numbers, translation in elements 13, 14, 15 — the placement the renderer draws the entity's geometry at, carrying the skew a decomposition into position / rotation / scale folds away. Also readable as `.transform.worldMatrix()`. - Bare `.scale` was removed — accessing it raises an error pointing callers at `localScale` / `lossyScale`. - **Sealed proxies**: every proxy errors on `__newindex` (sealing). Mutations go through `position.set` / `attribute.set` / etc., not direct field writes. - **Caching**: the entity-proxy cache is keyed by `id` with weak values (`__mode = "v"`), so dropped proxies are GC'd. Within one proxy, the `M_PREFIX`-keyed cache survives until the proxy itself is collected. - The `.component` sub-proxy is built by `__native_entity.component_subproxy` — Rust-side helper that constructs the ~15-closure component-proxy table once per entity. Subsequent `.component` accesses hit the cached entry.
▲ 0↑ born
▣
module · born here
❒asset
# value_type Converts a **handle-backed value type** — `ColorSequence`, `NumberSequence` — between the live object a session holds and the durable payload its `serialize()` produces. A sequence has two forms: | Form | Shape | Meaning | |------|-------|---------| | live | `{ kind = "ColorSequence", __h = 13 }` | `__h` names a curve in **this process's** curve registry | | durable | `{ kind = "ColorSequence", keypoints = { … } }` | the curve itself, which any session rebuilds | `__h` is minted when the curve is created and is meaningful for exactly as long as that registry lives, so a record carrying it names nothing in the session that reads it back. The keypoints are the authored data, and they cross every boundary intact — which is why a record carries the durable form and a live component field carries the live one. Both forms carry `kind`, so a value read out of a component field says which converter owns it whether it came from memory or from a record. The two are told apart by which of `__h` / `keypoints` the table itself holds. `Field.table` copies the table it is handed and drops its metatable, so a live value read back off a component field answers no methods at all. The field gives that same stored table back on every read, so `bind` puts the methods on it once and the field answers them from then on. ## API ```luau local ValueType = require("modules.value_type") local color = ColorSequence.new({ 1, 0.3, 0.1 }, { 0.1, 0.1, 1 }) ValueType.kindOf(color) -- "ColorSequence" ValueType.isLive(color) -- true ValueType.isPayload(color:serialize()) -- true -- The keypoints out of either form. The second argument is the kind to read a -- bare `{ __h = n }` as, for a caller that knows its field's type. ValueType.keypoints(component.color, "ColorSequence") -- The durable payload for a live value; nil for anything else. ValueType.serialize(color) -- { kind = "ColorSequence", keypoints = { … } } -- The live value a payload names, rebuilt in this session; nil for anything else. ValueType.revive(record.color) -- A whole component field map, with every payload rebuilt. local applyData, rebuilt = ValueType.withRevived(record.data) -- Put the methods back on a value read out of a component field. ValueType.bind(component.color, "ColorSequence"):evaluate(0) ``` `withRevived` returns the **same table** when the map carries no payload, so the common case allocates nothing; the second return says how many fields were rebuilt. ## Consumers `scene_saver` asks per field as it serialises an entity, and writes the payload in place of the handle. `scene_loader` rebuilds the payloads in a component's record before applying it, so the component takes the field holding the same curve it held when the record was written. `dirty_hot_reload` does the same as it applies a body, and reduces the live side of its record comparison to the durable form so both sides are compared in one shape.
▲ 0↑ born
▣
module · born here
❒asset
# component_proxy Computed-property registry shared by every component proxy. The proxy ITSELF is installed in Rust (`crates/zero_scripting/src/world_context.rs::setup_public_dirty_tracking`) — this module owns only the cross-VM-instance bookkeeping for `public.X = computed(fn)` declarations: the global registry table and the sentinel-string protocol that lets dispatch survive any serialization the engine performs on the public table. Installs `computed` as a `_G` global via `M.installGlobal()`. ## Exports - `M.computed(fn: (any) -> any) -> string` — stash `fn` keyed by a deterministic sentinel and return the sentinel string. Errors if `fn` is not a function. - `M.installGlobal()` — install `computed` as a `_G.computed` so component modules can write `public.X = computed(fn)` without an explicit require. Prelude calls this once. ## Usage ```luau local CP = require("@builtin::modules.component_proxy") -- Once, at boot (prelude does this): CP.installGlobal() -- In a component module's top-level body: public.area = computed(function(self) return self.w * self.h end) ``` ## Notes - The sentinel prefix (`__zero_computed__`) and registry global name (`__zero_computed_registry__`) are duplicated in `crates/zero_scripting/src/world_context.rs` (`COMPUTED_SENTINEL_PREFIX`). Renaming either side requires the matching rename on the other. - The registry is a `_G` table; it survives module hot-reload but every call to `computed(fn)` registers a fresh sentinel (sentinel includes `tostring(fn)`), so re-running a component module replaces its computed entries rather than leaking old ones — provided `fn` is a fresh closure each time. - `installGlobal` is idempotent. Re-running it just overwrites the global with the same function.
▲ 0↑ born
▣
module · born here
❒asset
# luau_introspect (module) Parses Luau/asset source TEXT into structured declarations. Pure string/pattern parsing (`string.find` / `string.match` / `string.gmatch`) over the literal source — no compiling, no VM spawn, no runtime state, no engine FFI. Given source text, it extracts what is written in the file. This is the shared parser every per-assetType `M.ref.inspect` hook uses to turn a component/module/tool's own source into the structured record `asset.inspect` returns. ## Exports - `docstrings(src) -> DocMap` — scans `--!desc` / `--!arg` / `--!return` / `--!example` comment runs and binds each run to the name of the declaration on the next non-comment line (the identifier after `function`, `public:`, `local function`, or `M.`). - `publicFields(src) -> { FieldEntry }` — parses `public = { name = Field.<kind>(default, mode), ... }` table-literal entries and module-scope `public.<name> = Field.<kind>(...)` assignments. - `events(src) -> { EventEntry }` — parses `events = { name = Event(payloadSchema?, syncMode?), ... }` table-literal entries. - `methods(src) -> { MethodEntry }` — parses `function public:<name>(...)` / `typed function public:<name>(...)` declarations (params + optional return type). - `lifecycleHooks(src, catalog) -> { string }` — top-level `function <name>(` declarations whose name is in the caller-supplied `catalog` array. - `moduleExports(src) -> { ExportEntry }` — parses `M.<name> = function` assignments, `function M.<name>(...)` declarations, and a trailing `return { foo = foo, ... }` export table. - `forSource(src, computeFn) -> any` — memoises `computeFn(src)` keyed on `src` itself in a module-local table. The same source text returns the cached result without re-running `computeFn`; different source text recomputes. Keyed on the source content (not an asset checksum) because composite/container assets (component folders, scene folders, etc.) all hash to the empty-string checksum — a checksum-keyed cache collapses every such asset's detail onto whichever one was inspected first. ## Usage ```luau local Introspect = require("modules.luau_introspect") local src = self:getInitScript() local detail = Introspect.forSource(src, function(s) return { fields = Introspect.publicFields(s), events = Introspect.events(s), methods = Introspect.methods(s), hooks = Introspect.lifecycleHooks(s, lifecycleCatalog), } end) ``` ## Notes - Every function is a pure string→table transform: same source text in, same structured table out, every time. No reads of live component state, no entity/world queries. - Structural scanning runs over a comment- and string-aware **mask** of the source (a same-length copy where comment bodies and string-literal contents are blanked to spaces, delimiters and newlines preserved). Pattern matches and brace/paren balancing run on the mask — so a brace, `Field.`, `Event(`, or a whole entry that appears inside a comment or a string literal never registers — while captured values are sliced from the original source at the same (aligned) indices. This makes the parser robust to apostrophes in comments, commented-out entries, string defaults containing `}` / `Field.` / `Event(`, and `[[ ]]` long strings. - `default` / `category` are raw trimmed source-text slices (with any trailing comment stripped), not evaluated values. - For a **ref constructor** (`Field.assetRef` / `dataRef` / `resource` / `componentRef`) the FIRST positional argument is the category / contract / componentType selector, captured under `category`; the SECOND argument is captured under `default`. - For every **other** Field kind the first argument is the `default` and `category` is nil. - A trailing `Sync` / `NoSync` identifier is captured as `sync`, tolerating a trailing comma or a trailing comment after the last argument. - Field/Event entries whose key is a computed or indirect expression (`[expr] = Field.number(...)`) are omitted — only literal identifier keys are recognised. - `methods` captures a `...` variadic parameter (as `name = "..."`), with its `: <type>` annotation when present. - A `--!desc` written above a field/event declaration (`<name> = Field.<kind>(...)` / `<name> = Event(...)`) or a `public.<name> =` assignment binds to that name in `docstrings`, and is folded into the corresponding `publicFields` / `events` entry's `desc`. - `lifecycleHooks` never hardcodes the lifecycle-callback catalog; the caller passes the list of names to match against. - `forSource`'s memo is bounded (oldest-key eviction) so a long session touching many distinct asset versions can't grow it without limit.
▲ 0↑ born
▣
module · born here
❒asset
# assetType.shared.ref Builds the metatable every `AssetRef` (the `{ __ref, type, name, guid, identity, path }` envelope returned by `asset.resolve` / `asset.ref`) carries. Replaces the per-call C metatable that used to live in `crates/zero_scripting/src/ffi/bindings/asset.rs::attach_asset_ref_metatable`. ## Public API ```lua local assetRef = require("@builtin::assetTypes.assetType.shared.ref") -- Called by the Rust factory (push_asset_ref_handle) via -- _G.__build_asset_ref_proxy after the envelope fields are set. assetRef.build(envelope) -- attaches AssetRefMT, returns the table ``` The module exposes no invalidation surface — discovery is registry-driven and hot-reload is transparent (see "Discovery and hot-reload" below). ## Method dispatch order `ref.foo` is resolved in this order: 1. **Default methods** — `getSource` / `getBytes` / `getText` / `exists` / `inspect` and the lazy `meta` property. Per-type definitions cannot override these (engine contract from gh#1889). 2. **Per-type `ref` table** — discovered through `asset.resolve(<typename>, "assetType")` + `require(<identity>.behavior)`. The lookup resolves to whichever `<typename>.assetType/` folder is currently registered for the typename (built-in, user, or library-vendored) and pulls its `ref` methods. 3. **Nil** — unknown key, same shape as the original C metatable. ## Adding methods for a new asset type Drop a `behavior.luau` into the type folder: ``` src/lua/lib/assetTypes/<typename>.assetType/ ├── type.yaml -- structural spec (existing) ├── behavior.luau -- ref + global behaviour (new) └── template/ -- placeholder body (existing) ``` `behavior.luau` returns: ```lua return { ref = { --!desc Per-instance method on every `<typename>` AssetRef. myMethod = function(self, ...) ... end, }, global = { -- Reserved for the asset.<typename>.* surface — not consumed -- yet; included so the contract from gh#1889 lands cleanly. }, } ``` Within `ref` methods, `self` is the AssetRef envelope: `self.path`, `self.identity`, `self.guid`, `self.type`, `self.name` are available. Engine-required fields (`guid`, `path`, `identity`, `type`) must not be redefined. ## Discovery and hot-reload Type discovery uses the asset registry, so it doesn't matter where a `<typename>.assetType/` folder lives — engine built-ins under `@builtin::assetTypes.<X>`, user types dropped under `/zero/source/<X>.assetType/`, or types vendored inside an imported library all resolve identically. The dispatcher walks `asset.resolve(<typename>, "assetType")` → canonical identity → `require <identity>.behavior` on every miss; the asset registry is the cache for the resolve half and Luau's `_LOADED` is the cache for the require half. Hot-reload is transparent. When a `<typename>.assetType/behavior.luau` is rewritten through VFS, the engine's `invalidate_require_cache_for_state` re-runs the module and mutates the cached `_LOADED[<identity>.behavior]` table in place. Dispatch reads `mod.ref` on every access (rather than caching a sub-pointer), so existing AssetRefs see the new methods on the very next access — no Rust-side hook into this module is needed, and no invalidation API is exposed for callers to call. The same registry-driven path also handles type **registration** correctness: dropping a new `<typename>.assetType/behavior.luau` updates the asset registry, the next `asset.resolve(<typename>, "assetType")` returns the new ref, and AssetRefs of that type immediately gain the new methods. ## Why Luau, not Rust The asset surface needed per-category behaviour (`material:getProperties()`, `bundle:instantiate(entityId)`, `tool:run(args)`) and the old shape couldn't grow without an FFI change per method. Owning the proxy in Luau means: - Type-specific methods ship as content (the `<typename>.assetType/` folder), not engine code. - Adding a new asset type drops in a new folder; no Rust rebuild. - Methods can call any other Luau global (`vfs`, `asset`, `Material`, `entity`, etc.) directly, instead of going through narrow FFI primitives. ## Consumers - `crates/zero_scripting/src/ffi/bindings/asset.rs::attach_asset_ref_metatable` — the Rust factory that pushes envelopes invokes `_G.__build_asset_ref_proxy` once per push. - Every Luau call site of `asset.resolve` / `asset.ref` / `entity(id).component.X.material` / any binding declared `AssetRef<...>` — they all receive envelopes built through this module.
▲ 0↑ born
▣
module · born here
❒asset
# yaml.module YAML decode + encode for authored engine content. Pure Luau, no engine dependencies — a general parsing library, not data-system plumbing. ```luau local Yaml = require("@builtin::modules.yaml") local doc = Yaml.decode(vfs.read(path)) vfs.write(path, Yaml.encode(doc)) ``` ## Supported subset - Block mappings (`key: value`) and block sequences (`- item`), nested to any depth. - Compact sequence-of-mapping entries (`- path: x` / ` kind: y`). - Flow collections — `{ a: 1, b: 2 }` and `[1, 2, 3]` — single line. - Plain scalars with type inference: `~` / `null` → nil, `true` / `false` → boolean, numeric literals → number, everything else → string. - Single- and double-quoted strings (`''` escapes a literal quote inside a single-quoted string; `\n` `\t` `\r` `\\` `\"` inside a double-quoted one). A quote opens a quoted scalar where a scalar begins — after `key:`, after a `- ` entry marker, or inside a flow collection — so a plain scalar carries an apostrophe as data (`note: the world's rules`). Quoted keys are read in flow mappings. - `#` comments — honored outside quotes, so a `#` inside a quoted string is kept as data. - Literal (`|`) and folded (`>`) block scalars, with a chomping indicator (`-` strips the final newline, `+` keeps every trailing blank line, none keeps one newline) and an indentation indicator (`1`-`9`: the content's indent past the parent's), in either order (`|-`, `>+`, `|2`, `>1-`). A block scalar's content is literal text: quotes (balanced or not), `#`, `key: value` and `---` inside it are data. - An optional leading `---` document marker. ## Loud-error contract Constructs outside the subset above raise instead of misparsing: anchors/aliases (`&`, `*`), tags (`!`), directives (`%`), multi-document streams (a second `---`), a block-scalar header it cannot read (`|x`, `|-+`, `|0`), tab indentation, and inconsistent indentation (both block structure and block-scalar bodies). Every error carries the offending line number (`yaml: line N: ...`), so a bad config points straight at the line to fix. Tab indentation is rejected everywhere — including block-scalar body lines whose leading whitespace contains a tab. A tab after the first non-space character of a line is content and passes through. ## API - `Yaml.decode(text: string): any` — decode a YAML document into a Luau value (table, scalar, or nil for an empty document). Raises on malformed input or an unsupported construct. - `Yaml.decodeWithComments(text: string): (any, boolean)` — decode as `decode` does, and report whether the text carries a comment by the decoder's own rule: a `#` at a line's start or after whitespace, outside a quoted scalar and outside a block scalar. Re-encoding the value drops such comments, so an editor that writes a document back reads this first. - `Yaml.encode(value: table): string` — encode a Luau table as YAML (block style, two-space indent, alphabetically sorted keys). Raises on values YAML can't represent (functions, userdata, non-string mapping keys). A top-level empty table encodes to an empty document, which decodes back to `nil` (the YAML empty-document ambiguity).
▲ 0↑ born
backing path · assetTypes/scene.assetType/shared.module/saver.module

Problems

Everything affecting this asset right now: its own problems, anything wrong inside it, and problems on its direct dependencies.

0problems
No problems reported on this asset or on the parts this page shows. Parts it leaves out are not counted here.
⌬ZeroMind agent review · awaiting first pass
Findings
Reviewer findings (handle · model · tag · quoted note) appear here once the per-pass review log lands. Today only the rolled-up agent_score is exposed.
usability—
did it work as advertised
quality—
authoring polish + cohesion
performance—
frame & memory budget held
agent review score
—
/ 100
awaiting first pass
usability × 0.40
+ quality × 0.35
+ performance × 0.25
± compat factor

Usability ratings

Did the part work as advertised when consumers tried to drop it in. Separate from upvotes: those are taste; this is "did it function".

—%no reports yet
Sign in to report whether this part worked for you.
Discussion

Scoped to this part · feeds back into the world's score.

0comments
Sign in to post.sign in
No comments yet. Be the first.