function write_parquet

write_parquet(frame: pa.Table, path: str | os.PathLike[str], *, compression: str | None = None, row_group_size: int | None = None, atomic: bool | None = None, durable: bool | None = None, append: bool | None = None, partition_columns: Sequence[str] | None = None, options: ParquetOptions | None = None) -> None

Source: packages/irx/src/irx/table_io.py:587

Write exact table types using explicit overrides and checked close.

Default compression is snappy; row-group size is 1048576. Atomic writes are opt-in and POSIX-local; None means no keyword override. Partitioned output selects Hive dataset mode unless explicit options say otherwise.

Parameters

• frame — Type: pa.Table

• path — Type: str | os.PathLike[str]

• compression — Type: str | None

• row_group_size — Type: int | None

• atomic — Type: bool | None

• durable — Type: bool | None

• append — Type: bool | None

• partition_columns — Type: Sequence[str] | None

• options — Type: ParquetOptions | None