Struct EvaluationSupervisor 

pub struct EvaluationSupervisor { /* private fields */ }

Connects an append-only fact journal to a host policy, evidence reader, and auxiliary-run service. The policy is the only component that decides when an evaluation is useful; Core only enforces identity, bounds, and cleanup.

Implementations

impl EvaluationSupervisor

pub fn new( journal: Arc<dyn ExecutionFactJournal>, reader: Arc<dyn EvidenceReader>, auxiliary: Arc<dyn AuxiliaryRunService>, policy: Arc<dyn EvaluationPolicy>, ) -> Self

pub fn with_dispatch_ledger( journal: Arc<dyn ExecutionFactJournal>, reader: Arc<dyn EvidenceReader>, auxiliary: Arc<dyn AuxiliaryRunService>, policy: Arc<dyn EvaluationPolicy>, dispatch_ledger: Arc<dyn EvaluationDispatchLedger>, ) -> Self

Construct a supervisor whose dispatch claims survive process restart. The ledger is still only a replay/lease mechanism; result persistence and host authorization remain separate concerns.

pub fn cancellation(&self) -> CancellationToken

pub fn cancel(&self)

pub async fn shutdown(&self)

Cancel all admitted auxiliary work and release this supervisor’s in-memory admission history. Auxiliary services still own their terminal snapshots; the parent cancellation token makes cooperative executors settle without blocking the caller.

pub async fn pending_count(&self) -> usize

Return the number of auxiliary runs currently admitted but not yet terminal. This is an observation only and never grants admission.

pub async fn observe_event( &self, frame: ExecutionFrameV1, record: &RunEventRecord, ) -> Result<EvaluationDispatch, SupervisorError>

Record one runtime event and, if the host policy selects the matching boundary, dispatch one isolated auxiliary run. The returned handle is advisory: the parent execution is never implicitly blocked by the auxiliary result.

Trait Implementations

impl Debug for EvaluationSupervisor

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

Formats the value using the given formatter.