Struct RealtimeScheduler 

pub struct RealtimeScheduler<B: RealtimeBackend> { /* private fields */ }

Fair bounded realtime scheduler generic over the selected backend.

Implementations

impl<B: RealtimeBackend> RealtimeScheduler<B>

pub fn new( model: &RealtimeModel<B>, limits: SchedulerLimits, ) -> Result<Self, RealtimeError<B::Error>>

Binds an empty scheduler to one selected model.

pub fn register_request( &mut self, model: &RealtimeModel<B>, request: RequestId, sampling: RealtimeSampling, ) -> Result<(), RealtimeError<B::Error>>

Registers a request with fresh backend-owned state.

pub fn register_request_with_session( &mut self, model: &RealtimeModel<B>, request: RequestId, session: RealtimeSession<B>, ) -> Result<(), RealtimeError<B::Error>>

Registers a previously released request session.

pub fn enqueue( &mut self, model: &RealtimeModel<B>, request: RequestId, input: B::Input, ) -> Result<WorkId, RealtimeError<B::Error>>

Enqueues one encoded or forced frame.

pub fn enqueue_with_deadline( &mut self, model: &RealtimeModel<B>, request: RequestId, input: B::Input, deadline: Option<Instant>, ) -> Result<WorkId, RealtimeError<B::Error>>

Enqueues one frame with an optional absolute deadline.

pub fn enqueue_batch( &mut self, model: &RealtimeModel<B>, request: RequestId, inputs: Vec<B::Input>, ) -> Result<Vec<WorkId>, RealtimeError<B::Error>>

Atomically enqueues ordered frames.

pub fn run_queued( &mut self, model: &mut RealtimeModel<B>, ) -> Result<Vec<RealtimeCompletedStep<B::Output>>, RealtimeError<B::Error>>

Advances one unbounded fair scheduling turn.

pub fn run_bounded( &mut self, model: &mut RealtimeModel<B>, max_frames: usize, ) -> Result<Vec<RealtimeCompletedStep<B::Output>>, RealtimeError<B::Error>>

Advances at most max_frames fair-ordered transitions.

pub fn finish_request( &mut self, request: RequestId, ) -> Result<(), RealtimeError<B::Error>>

Completes one request and drops its backend session.

pub fn cancel_request( &mut self, request: RequestId, ) -> Result<(), RealtimeError<B::Error>>

Cancels one request and discards queued frames.

pub fn release_request( &mut self, request: RequestId, ) -> Result<RealtimeSession<B>, RealtimeError<B::Error>>

Releases an idle request for persistence or resumption.

pub fn forget_terminal_request( &mut self, request: RequestId, ) -> Result<RequestStatus, RealtimeError<B::Error>>

Removes a terminal identity for explicit reuse.

pub fn request_status(&self, request: RequestId) -> Option<RequestStatus>

Lifecycle state for a known request.

pub fn queued_for_request(&self, request: RequestId) -> usize

Queued frame count for one request.

pub fn set_request_sampling( &mut self, model: &RealtimeModel<B>, request: RequestId, sampling: RealtimeSampling, ) -> Result<(), RealtimeError<B::Error>>

Replaces sampling controls for an idle active request.

pub fn report(&self) -> SchedulerReport

Generic occupancy and lifecycle telemetry.

pub fn capabilities(&self) -> SchedulerCapabilities

Configured bounds and observed backend capabilities.