Enum BatchPolicy 

pub enum BatchPolicy {
    Fixed(usize),
    Adaptive,
}

How aggressively the owner drains the remote-free queue.

Variants

Fixed(usize)

Drain every N remote frees. v0.1 default = 64.

Adaptive

Stepped-threshold adaptive policy. The drain threshold steps through 5 levels — [8, 16, 32, 64, 128] — based on observed queue depth relative to queue_capacity:

• Queue length > 75% of capacity → step down (smaller batch, drain sooner) to relieve back-pressure on remote senders.
• Queue length < 25% of capacity → step up (larger batch, drain less often) to amortize the drain cost across more frees.

A cooldown of ADAPTIVE_COOLDOWN_TICKS maybe_drain calls between adjustments prevents oscillation. Initial step is 3 (threshold = 64) — matches the Fixed(64) v0.1 default so an Adaptive-policy owner behaves like a Fixed(64) owner until the workload pushes the threshold off the middle level.

All arithmetic is integer-only; the 0.25 / 0.75 hysteresis band is encoded as q*4 < cap / q*4 > cap*3. No floating point.

This is the v1.0 control law; the v2.0 EMA-based upgrade is gated on benchmark validation against this baseline.

Trait Implementations

impl Clone for BatchPolicy

fn clone(&self) -> BatchPolicy

Returns a duplicate of the value.

fn clone_from(&mut self, source: &Self)

Performs copy-assignment from source.

impl Copy for BatchPolicy

impl Debug for BatchPolicy

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

Formats the value using the given formatter.

impl Default for BatchPolicy

fn default() -> Self

Returns the “default value” for a type.