DataType

Mirrors Spark’s DataType hierarchy for type-safe schema operations.

See

Spark types live in org.apache.spark.sql.types. The main mapping constraint in JavaScript is that number is IEEE-754 float64, so it cannot represent every Spark integer value exactly.

Spark type	Size	JavaScript representation
`ByteType`	1 byte	`number`
`ShortType`	2 bytes	`number`
`IntegerType`	4 bytes	`number`
`LongType`	8 bytes	`bigint` when precision matters
`FloatType`	4 bytes	`number`
`DoubleType`	8 bytes	`number`
`DecimalType`	variable	string-like decimal encoding
`StringType`	variable	`string`
`BooleanType`	1 bit	`boolean`
`BinaryType`	variable	`Uint8Array`
`TimestampType`	8 bytes	`Date` with microsecond precision loss
`DateType`	4 bytes	`Date`

Arrow’s type system is close enough to Spark’s SQL types that it can carry result batches efficiently with very little translation on the client side.