Struct TensorFlowController 

pub struct TensorFlowController { /* private fields */ }

Controls the flow of tensor data through a processing pipeline.

The internal queue is kept sorted by priority descending (Critical first), then by enqueued_at_tick ascending (FIFO within a priority level).

Implementations

impl TensorFlowController

pub fn new(config: FlowControllerConfig) -> Self

Create a new controller with the given configuration, starting in Running state.

pub fn admit(&mut self, item: FlowItem) -> bool

Attempt to admit a new item into the queue.

Returns true if the item was admitted, false if it was dropped.

pub fn process_tick(&mut self) -> Vec<FlowItem>

Process one tick: drain items from the front of the queue up to the byte budget.

Returns the items that were processed during this tick.

pub fn pause(&mut self)

Halt all flow immediately.

pub fn resume(&mut self)

Resume from a Paused state. The new state is determined by the current queue fill.

pub fn drain(&mut self)

Stop admitting new items; allow the queue to drain.

pub fn queue_len(&self) -> usize

Current number of items waiting in the queue.

pub fn stats(&self) -> &FlowStats

Reference to the current statistics.