Struct OperationContext 

pub struct OperationContext { /* private fields */ }

Cancellation, deadline, and task-class metadata for one bounded operation.

An operation is not a lifecycle owner and does not create a child TaskGroup. Tasks spawned for the operation remain registered with a fixed component owner while this context supplies operation-local cancellation and bounded join semantics.

Implementations

impl OperationContext

pub async fn wait_with_policy( &self, owner: &TaskGroup, policy: OperationWaitPolicy, ) -> RuntimeResult<OperationWaitOutcome>

Closes operation admission and asynchronously waits using explicit deadlines.

Tasks unfinished at the graceful deadline are asked to abort. The second deadline bounds confirmation of future destruction and owner settlement. An unconfirmed outcome leaves those tasks visible to the owner’s report. Deadlines bound cooperative waiting; blocked executor threads can delay polling and this method does not promise a hard wall-clock return time.

Errors

Returns an error without closing admission if owner is not this operation’s bound component owner.

pub fn new(deadline: Instant, task_kind: TaskKind) -> Self

Creates an operation with an absolute deadline.

pub fn without_deadline(task_kind: TaskKind) -> Self

Creates an operation that remains active until explicitly cancelled.

pub fn cancellation_token(&self) -> CancellationToken

Returns the operation-local cancellation token.

pub fn deadline(&self) -> Option<Instant>

Returns the absolute deadline, when one was configured.

pub fn task_kind(&self) -> TaskKind

Returns the task classification applied to operation tasks.

pub fn with_task_kind(&self, task_kind: TaskKind) -> Self

Returns a view of this operation with a different task classification.

Cancellation, deadline, owner binding, and active-task tracking remain shared with the original context.

pub fn is_cancelled(&self) -> bool

Returns whether cancellation has been requested.

pub fn cancel(&self)

Requests cancellation for every task in this operation.

pub fn close_admission(&self)

Closes task admission without cancelling tasks that were already accepted.

pub fn active_task_count(&self) -> usize

Returns the number of operation tasks that have not completed.

pub fn outcomes(&self) -> [(OperationOutcome, u64); 6]

Reads counters without retaining individual task history.

Counters share this operation’s lifetime and are individually atomic; completions can occur between fields in this snapshot.

pub fn set_outcome_observer( &self, observer: Arc<dyn OperationOutcomeObserver>, ) -> RuntimeResult<()>

Binds an observer before the first submission attempt.

Errors

Returns an error after owner binding or an earlier observer installation.

pub async fn cancel_and_wait( &self, owner: &TaskGroup, timeout: Duration, ) -> RuntimeResult<bool>

Cancels this operation and waits for all tasks registered with owner.

Returns true when every task completed before the shared timeout. Tasks still running at the deadline are aborted and awaited for a confirmation window of at most one second beyond it. false does not distinguish a confirmed abort from work whose destruction is still unconfirmed. Use Self::wait_with_policy after Self::cancel for explicit deadlines and completion evidence. Timers require the runtime to make progress; they are not a hard wall-clock return guarantee.

Errors

Returns an error when owner differs from the component owner used to spawn the operation.

pub async fn cancel_and_wait_until( &self, owner: &TaskGroup, deadline: ShutdownDeadline, ) -> RuntimeResult<bool>

Cancels this operation and bounds its asynchronous wait by deadline.

Behaves like Self::wait_until after requesting cancellation.

Errors

Returns an error when owner differs from the component owner used to spawn the operation.

pub async fn wait( &self, owner: &TaskGroup, timeout: Duration, ) -> RuntimeResult<bool>

Waits until no operation task is active and owner has settled every one, without requesting cancellation. Tasks still running at the deadline are aborted and awaited for a confirmation window of at most one second, so the call can return up to one second after timeout.

Tasks accepted during the wait are waited for too; close admission first when new tasks may still be submitted. Returns true when every task finished before the deadline; owner then no longer lists any of them. false does not confirm destruction. Use Self::wait_until to avoid an extra confirmation allowance, or Self::wait_with_policy to distinguish confirmed aborts from unconfirmed work.

Errors

Returns an error when owner differs from the component owner used to spawn the operation.

pub async fn wait_until( &self, owner: &TaskGroup, deadline: ShutdownDeadline, ) -> RuntimeResult<bool>

Waits for this operation’s tasks with no allowance beyond deadline.

Tasks still running at the deadline are asked to abort and the call returns false at once, without confirming that their futures were dropped. They stay registered with owner until they are, so the owner’s shutdown report accounts for any that remain. This matches TaskGroup::shutdown_until, which also does not extend its deadline. Runtime starvation or synchronous destructors can delay timer polling.

Errors

Returns an error when owner differs from the component owner used to spawn the operation.

Trait Implementations

impl Clone for OperationContext

fn clone(&self) -> Self

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Debug for OperationContext

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

Formats the value using the given formatter.