Struct Tier1SimdExecutor 

pub struct Tier1SimdExecutor { /* private fields */ }

Per-fiber Tier-1 executor. The scalar and native register kernels share no mutable state with another fiber.

Implementations

impl Tier1SimdExecutor

pub fn descriptor(&self) -> &Tier1SimdDescriptor

The plan’s description.

pub const fn stats(&self) -> Tier1SimdStats

The counters so far.

pub fn scalar_program(&self) -> &Arc<PolydatProgram>

The scalar program the plan falls back to.

pub const fn has_active_lease(&self) -> bool

Whether a lease is open.

pub fn remaining(&self) -> u64

Values left in the open lease, or zero.

pub fn set_broadcast( &mut self, name: &str, value: Value, ) -> Result<(), Tier1SimdError>

Freeze one scope-stable input value for subsequent packet evaluation. Rebinding during an owned lease is rejected rather than invalidating already-reserved logical inputs.

pub fn begin_lease( &mut self, lease: OrdinalBatchLease, activation_epoch: u64, ) -> Result<(), Tier1LeaseStartError>

Accept an owned source reservation. A rejection returns the lease in Tier1LeaseStartError so the caller can scalar-process it.

pub fn force_scalar_recovery(&mut self)

Discard an uncommitted register packet and process the remainder of the active lease through the retained scalar graph. The consumer frontier does not move and the shared source cursor is not touched.

pub fn drain_into(&mut self, output: &mut [Value]) -> usize

Drain committed scalar values in ordinal order. Calls may use arbitrary burst sizes; a partly consumed register packet remains in BatchState.