Function concatRequiredField

concatRequiredField<A extends ConcatTypedArray, P>(
    parts: P[],
    get: (p: P) => A,
    countOf: (p: P) => number,
    perItem?: number,
    label?: string,
    levelOf?: (index: number) => number,
): A
Concatenate a required typed-array field across parts.

Allocates a fresh array of sum(countOf(part)) * perItem elements (using the first part's constructor, so the dtype is preserved) and copies each part in order.

LADDER-DTYPE CONTRACT: every part must carry the field with the SAME typed-array dtype. TypedArray.set converts by VALUE, not by semantics — copying a later Uint8Array level (0..255 normalized) into a Float32Array output (0..1) would silently write 255× values, and Float32 into Uint8 truncates to garbage. The Python writer always emits one dtype per field across a ladder, so a mismatch means malformed data; fail fast with a descriptive error instead of rendering corruption.
Type Parameters
- A extends ConcatTypedArray
- P
Parameters
- parts: P[]
  Per-LOD data parts (assumed length ≥ 1).
- get: (p: P) => A
  Extract the field from a part.
- countOf: (p: P) => number
  Element count of a part (rows; multiplied by perItem).
- perItem: number = 1
  Components per element (e.g. 3 for positions, 1 for widths).
- label: string = 'field'
  Field name for the dtype-mismatch error message.
- levelOf: (index: number) => number = ...
  Maps a parts index back to its LADDER level number, for the error message only. Identity unless the caller passed a filtered subset (see concatOptionalField's zero-row abstainers) — naming the wrong level in a corrupt-store diagnostic is worse than not naming one.
Returns A
- Defined in src/data/loaders/progressive/concat-helpers.ts:50

Function concatRequiredField

Type Parameters

Parameters

Returns A

Settings