Struct BufferedBatch 

pub(super) struct BufferedBatch {
    pub batch: BufferedBatchState,
    pub range: Range<usize>,
    pub join_arrays: Vec<ArrayRef>,
    pub null_joined: Vec<usize>,
    pub size_estimation: usize,
    pub join_filter_not_matched_map: HashMap<u64, bool>,
    pub num_rows: usize,
}

A buffered batch that contains contiguous rows with same join key

BufferedBatch can exist as either an in-memory RecordBatch or a RefCountedTempFile on disk.

Fields

batch: BufferedBatchState

Represents in memory or spilled record batch

range: Range<usize>

The range in which the rows share the same join key

join_arrays: Vec<ArrayRef>

Array refs of the join key

null_joined: Vec<usize>

Buffered joined index (null joining buffered)

size_estimation: usize

Size estimation used for reserving / releasing memory

join_filter_not_matched_map: HashMap<u64, bool>

The indices of buffered batch that the join filter doesn’t satisfy. This is a map between right row index and a boolean value indicating whether all joined row of the right row does not satisfy the filter . When dequeuing the buffered batch, we need to produce null joined rows for these indices.

num_rows: usize

Current buffered batch number of rows. Equal to batch.num_rows() but if batch is spilled to disk this property is preferable and less expensive

Implementations

impl BufferedBatch

fn new( batch: RecordBatch, range: Range<usize>, on_column: &[PhysicalExprRef], ) -> Self

Trait Implementations

impl Debug for BufferedBatch

fn fmt(&self, f: &mut Formatter<'_>) -> Result

Formats the value using the given formatter.