Function write_data_files

fn write_data_files<G>(
    name: String,
    input: VecCollection<G, (Option<Row>, DiffPair<Row>), Diff>,
    batch_desc_input: &Stream<G, (Antichain<Timestamp>, Antichain<Timestamp>)>,
    connection: IcebergSinkConnection,
    storage_configuration: StorageConfiguration,
    materialize_arrow_schema: Arc<Schema>,
) -> (Stream<G, BoundedDataFile>, Stream<G, HealthStatusMessage>, PressOnDropButton)
where
    G: Scope<Timestamp = Timestamp>,

Write rows into Parquet data files bounded by batch descriptions. Rows are matched to batches by timestamp; if a batch description hasn’t arrived yet, rows are stashed until it does. This allows batches to be minted ahead of data arrival.