Struct MorselWorkerPool 

pub struct MorselWorkerPool { /* private fields */ }

Query-level bounded worker pool with morsel-style dynamic task assignment.

Workers are created once per query and persist across multiple Exchange node invocations. Partition tasks are dynamically claimed from a shared atomic counter, so faster workers naturally process more work — this eliminates the static round-robin load imbalance of per-Gather threads.

Implementations

impl MorselWorkerPool

pub fn new(max_workers: usize) -> Self

Create a new pool with max_workers persistent worker threads.

pub fn max_workers(&self) -> usize

Number of workers in this pool.

pub fn shutdown(&mut self)

Signal all workers to stop and wait for them to exit.

Must be called from a non-worker thread — never from within a worker that holds the final runtime reference.

pub fn shutdown_shared(&self)

Signal all workers to stop through a shared reference.

Sets the stop flag and closes the batch channel. Workers blocked on recv observe the closed channel and exit; the stop flag covers workers between batches. Safe to call while queries still hold runtime references to this pool.

pub fn detach(&mut self)

Signal workers to stop without joining.

Safe to call from any context (including a worker that may own the last Arc<ExecutionRuntime>). Workers exit naturally when they observe the stop flag or the closed channel.

Trait Implementations

impl Debug for MorselWorkerPool

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

Formats the value using the given formatter.

impl Drop for MorselWorkerPool

fn drop(&mut self)

Executes the destructor for this type.

fn pin_drop(self: Pin<&mut Self>)

🔬This is a nightly-only experimental API. (pin_ergonomics)

Execute the destructor for this type, but different to Drop::drop, it requires self to be pinned.

impl TaskScheduler for MorselWorkerPool

fn submit(&self, batch: Arc<PartitionBatch>)

Submit a batch of partition tasks for execution.

fn max_workers(&self) -> usize

Number of worker threads in this scheduler.