Struct HistoryCursor 

pub struct HistoryCursor { /* private fields */ }

Ordered cursor over a workflow’s recorded history.

Commands consume only their own events. Asynchronous arrivals (signals, child terminals, a parallel activity’s events) can be recorded anywhere inside another command’s event range, so consumption is tracked per event: a resolved command marks exactly its own events consumed and skipped interior events stay matchable for their own commands. position is the low-water mark — the first index that is neither consumed nor already skipped past — and scans resume from it.

Implementations

impl HistoryCursor

pub fn new(events: Vec<Event>) -> Result<Self, DurabilityError>

Builds a cursor from ordered read-history output.

Errors

Returns DurabilityError::HistoryShape when event sequence numbers decrease.

pub fn current_sequence(&self) -> Option<u64>

Returns the sequence number at the current cursor position, or None when exhausted.

pub fn events(&self) -> &[Event]

Returns the ordered history backing this cursor.

pub const fn position_index(&self) -> usize

Returns the current zero-based index into the owned history.

pub fn next_key(&self, family: RecordedEventFamily) -> Option<CorrelationKey>

Returns the next matchable correlation key for family without consuming history.

pub fn fast_forward_to_key(&mut self, key: &CorrelationKey)

Advance past recorded commands that earlier resolver instances of the same live execution already consumed.

NIF calls each build a fresh resolver whose cursor starts at the top of history, so commands resolved by earlier calls sit before the one being resolved now. Skipping stops at the first matchable entry whose correlation key equals key — or at history end, leaving live resolution to proceed — so a genuinely out-of-order command for an already-positioned key still surfaces as a mismatch in HistoryCursor::resolve_next. Strict full-history replay (recovery, the replay driver) never calls this.

pub fn fast_forward_to_child_terminal(&mut self, child_workflow_id: &WorkflowId)

Advance to the recorded terminal outcome for child_workflow_id.

AwaitChild has no positional correlation key — its replay identity is the child workflow id returned by the matching spawn — so it gets the same skip treatment keyed commands receive from HistoryCursor::fast_forward_to_key: recorded commands consumed by earlier resolver instances of the same live execution are skipped until the awaited child’s ChildWorkflowCompleted/ChildWorkflowFailed is reached. With no recorded terminal for that child the cursor exhausts, leaving resolution to hand off live. Strict full-history replay (recovery, the replay driver) never calls this.

pub fn resolve_child_terminal( &mut self, child_workflow_id: &WorkflowId, ) -> ChildTerminalResolveResult

Resolves the next recorded child terminal outcome for child_workflow_id.

pub fn resolve_next( &mut self, family: RecordedEventFamily, expected_key: CorrelationKey, ) -> CursorResolveResult

Resolves the next recorded outcome for the expected family and correlation key.

Trait Implementations

impl Clone for HistoryCursor

fn clone(&self) -> HistoryCursor

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Debug for HistoryCursor

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

Formats the value using the given formatter.