Struct JoinLeftData 

pub(super) struct JoinLeftData {
    pub(super) hash_map: Box<dyn JoinHashMapType>,
    batch: RecordBatch,
    values: Vec<ArrayRef>,
    visited_indices_bitmap: Mutex<BooleanBufferBuilder>,
    probe_threads_counter: AtomicUsize,
    _reservation: MemoryReservation,
    pub(super) bounds: Option<Vec<ColumnBounds>>,
}

HashTable and input data for the left (build side) of a join

Fields

hash_map: Box<dyn JoinHashMapType>

The hash table with indices into batch

batch: RecordBatch

The input rows for the build side

values: Vec<ArrayRef>

The build side on expressions values

visited_indices_bitmap: Mutex<BooleanBufferBuilder>

Shared bitmap builder for visited left indices

probe_threads_counter: AtomicUsize

Counter of running probe-threads, potentially able to update visited_indices_bitmap

_reservation: MemoryReservation

We need to keep this field to maintain accurate memory accounting, even though we don’t directly use it. Without holding onto this reservation, the recorded memory usage would become inconsistent with actual usage. This could hide potential out-of-memory issues, especially when upstream operators increase their memory consumption. The MemoryReservation ensures proper tracking of memory resources throughout the join operation’s lifecycle.

bounds: Option<Vec<ColumnBounds>>

Bounds computed from the build side for dynamic filter pushdown

Implementations

impl JoinLeftData

pub(super) fn new( hash_map: Box<dyn JoinHashMapType>, batch: RecordBatch, values: Vec<ArrayRef>, visited_indices_bitmap: Mutex<BooleanBufferBuilder>, probe_threads_counter: AtomicUsize, reservation: MemoryReservation, bounds: Option<Vec<ColumnBounds>>, ) -> Self

Create a new JoinLeftData from its parts

pub(super) fn hash_map(&self) -> &dyn JoinHashMapType

return a reference to the hash map

pub(super) fn batch(&self) -> &RecordBatch

returns a reference to the build side batch

pub(super) fn values(&self) -> &[ArrayRef]

returns a reference to the build side expressions values

pub(super) fn visited_indices_bitmap(&self) -> &Mutex<BooleanBufferBuilder>

returns a reference to the visited indices bitmap

pub(super) fn report_probe_completed(&self) -> bool

Decrements the counter of running threads, and returns true if caller is the last running thread