PyArrow is supported as an efficient binary format (instead of json) to write data to a Data Frame Table through pa.RecordBatch()
nisystemlink-clients only support the PyArrow format for the client.append_table_data() but still require the user to define the table format manually and the data query methods also return in non-PyArrow formats.
When a user has data in PyArrow format, it is highly likely that the user expects a workflow to infer or replicate the columns and types from the PyArrow schema for consistency.
PyArrow is supported as an efficient binary format (instead of json) to write data to a Data Frame Table through
pa.RecordBatch()nisystemlink-clientsonly support the PyArrow format for theclient.append_table_data()but still require the user to define the table format manually and the data query methods also return in non-PyArrow formats.When a user has data in PyArrow format, it is highly likely that the user expects a workflow to infer or replicate the columns and types from the PyArrow schema for consistency.