Struct WorkGraphMachine 

pub struct WorkGraphMachine;

Implementations

impl WorkGraphMachine

pub fn validate_item_projection(item: &WorkItem) -> Result<(), WorkGraphError>

Mechanically assert that a WorkItem’s projected lifecycle fields agree with its machine-owned machine_state authority.

This is a pure structural check: it borrows item immutably and can only return Ok(()) (every projection matches the machine state) or an Err rejecting the drift. It never synthesizes, repairs, or derives any lifecycle/revision field — the canonical truth lives in WorkGraphMachineState, and this guard merely rejects projections that disagree with it.

pub fn public_error_class( error: &WorkGraphError, ) -> Result<WorkGraphPublicErrorClass, WorkGraphError>

Resolve the public error class for a WorkGraphError.

The shell performs only a pure typed extraction of the error variant into a WorkGraphErrorKind discriminant (one kind per variant, no grouping). The variant->class POLICY is owned by the canonical WorkGraphLifecycleMachine: this drives the machine’s ClassifyWorkGraphPublicError input and mirrors the emitted WorkGraphPublicErrorClassified effect. The shell decides nothing.

pub fn classify_public_confirmation_admission( completion_policy: &WorkCompletionPolicy, ) -> Result<WorkPublicConfirmationAdmissionKind, WorkGraphError>

Resolve whether an untrusted PUBLIC caller may confirm an item with the given machine-owned completion policy.

The trust-scoped eligibility “only a self-attested completion policy may be confirmed by a public caller; every other policy requires trusted in-process host authority” is owned by the canonical WorkGraphLifecycleMachine, not the public-confirm surface. The shell performs only a pure typed extraction of the machine-owned completion_policy into the DSL observation, drives the machine’s ClassifyPublicConfirmationAdmission input, and mirrors the emitted PublicConfirmationAdmissionClassified verdict. The shell decides nothing and fails closed if the machine refuses or emits no verdict.

pub fn classify_create_completion_policy_admission( completion_policy: &WorkCompletionPolicy, ) -> Result<WorkCreateCompletionPolicyAdmissionKind, WorkGraphError>

Resolve whether a requested completion policy is admissible at CREATE for a non-goal work item.

The creation policy “non-goal work items must use the self-attest completion policy” is owned by the canonical WorkGraphLifecycleMachine, not the create shell. The shell performs only a pure typed extraction of the requested completion policy into the DSL observation, drives the machine’s ClassifyCreateCompletionPolicyAdmission input over a fresh authority, and mirrors the emitted CreateCompletionPolicyAdmissionClassified verdict. The shell decides nothing and fails closed if the machine refuses or emits no verdict.

pub fn classify_completion_policy_mutation_admission( item: &WorkItem, requested: &WorkCompletionPolicy, ) -> Result<WorkCompletionPolicyMutationAdmissionKind, WorkGraphError>

Resolve whether a requested completion-policy mutation is admissible.

The immutability invariant “a work item’s completion policy is fixed at creation and cannot be changed by an update” is owned by the canonical WorkGraphLifecycleMachine, not the shell. The shell performs only a pure typed extraction of the requested completion policy into the DSL observation (variant plus supervisor owner key plus reviewer quorum threshold), drives the machine’s ClassifyCompletionPolicyMutationAdmission input over the item’s recovered machine state, and mirrors the emitted CompletionPolicyMutationAdmissionClassified verdict. The machine compares the requested policy — in full — against its own machine-owned completion policy; this function decides nothing and fails closed if the machine refuses or emits no verdict.

pub fn classify_confirmation_admission( completion_policy: &WorkCompletionPolicy, requested_principal: Option<&WorkOwnerKey>, supplied_evidence_kind: WorkConfirmationEvidenceObservation, ) -> Result<WorkConfirmationAdmissionKind, WorkGraphError>

Resolve whether a trusted-path goal confirmation is admissible for a work item’s machine-owned completion policy.

The eligibility “is this confirming principal + supplied evidence kind admissible for this completion policy” is owned by the canonical WorkGraphLifecycleMachine, not the goal-confirm shell. The shell performs only pure typed extraction of the observations (the machine-owned completion policy + its supervisor owner key, the requested confirming principal owner key + kind, and the typed evidence-kind observation projected from the evidence’s typed confirmation classification), drives the machine’s ClassifyConfirmationAdmission input, and mirrors the emitted ConfirmationAdmissionClassified verdict. This function decides nothing and fails closed if the machine refuses or emits no verdict.

pub fn classify_terminality(item: &WorkItem) -> Result<bool, WorkGraphError>

Resolve whether a work item is terminal.

The shell extracts no fact: it drives the canonical WorkGraphLifecycleMachine’s ClassifyTerminality input over the item’s recovered machine state. The machine owns the lifecycle_phase and the terminality verdict (which phases are terminal); this function mirrors the emitted WorkItemTerminalityClassified.terminal, failing closed if the machine refuses or emits no verdict. It decides nothing.

pub fn classify_blocker_satisfied( gated_item: &WorkItem, blocker: Option<&WorkItem>, ) -> Result<bool, WorkGraphError>

Resolve whether a single blocking edge is satisfied.

The shell extracts only the raw blocker lifecycle phase (a pure observation projected from the blocker’s own machine state) and whether the blocker was resolvable at all, then drives the canonical WorkGraphLifecycleMachine’s ClassifyBlockerSatisfied input over the gated item’s recovered state. The machine owns the satisfaction POLICY (a blocking edge is satisfied iff its blocker reached terminal SUCCESS, Completed) and emits the verdict; this function mirrors it. The caller mechanically fans-in (counts) the unsatisfied edges. Fails closed.

pub fn create_item( request: CreateWorkItemRequest, realm_id: String, namespace: WorkNamespace, now: DateTime<Utc>, ) -> Result<(WorkItem, WorkGraphEvent), WorkGraphError>

pub fn update_item( item: WorkItem, request: UpdateWorkItemRequest, now: DateTime<Utc>, ) -> Result<(WorkItem, WorkGraphEvent), WorkGraphError>

pub fn claim_item( item: WorkItem, request: ClaimWorkItemRequest, now: DateTime<Utc>, ) -> Result<(WorkItem, WorkGraphEvent), WorkGraphError>

pub fn claim_ready_item( item: WorkItem, request: ClaimWorkItemRequest, now: DateTime<Utc>, ) -> Result<(WorkItem, WorkGraphEvent), WorkGraphError>

pub fn refresh_eligibility( item: WorkItem, unresolved_blocker_count: u64, now: DateTime<Utc>, ) -> Result<Option<(WorkItem, WorkGraphEvent)>, WorkGraphError>

pub fn release_item( item: WorkItem, request: ReleaseWorkItemRequest, now: DateTime<Utc>, ) -> Result<(WorkItem, WorkGraphEvent), WorkGraphError>

pub fn block_item( item: WorkItem, expected_revision: u64, now: DateTime<Utc>, ) -> Result<(WorkItem, WorkGraphEvent), WorkGraphError>

pub fn close_item( item: WorkItem, request: CloseWorkItemRequest, now: DateTime<Utc>, ) -> Result<(WorkItem, WorkGraphEvent), WorkGraphError>

pub fn add_evidence( item: WorkItem, request: AddEvidenceRequest, now: DateTime<Utc>, ) -> Result<(WorkItem, WorkGraphEvent), WorkGraphError>

pub fn classify_readiness( item: &WorkItem, now: DateTime<Utc>, ) -> Result<bool, WorkGraphError>

Classify whether a work item is ready to claim at now.

The shell extracts only the raw wall-clock now (a pure observation) and drives the canonical WorkGraphLifecycleMachine’s ClassifyReadiness input over the item’s recovered machine state. The machine owns the readiness POLICY — reproducing exactly the Claim transition guards (ClaimOpen / ClaimExpiredInProgress) — and emits the verdict; this function mirrors the emitted WorkItemReadinessClassified.ready, failing closed if the machine refuses or emits no verdict. It decides nothing.

pub fn is_ready(item: &WorkItem, now: DateTime<Utc>) -> bool

Mirror the machine-owned readiness verdict as a bool, failing closed.

The readiness verdict belongs to the canonical WorkGraphLifecycleMachine (see WorkGraphMachine::classify_readiness). This filter-facing projection mirrors the machine bool; if the machine refuses to classify, it fails closed by treating the item as NOT ready so an unclassifiable item is never offered as claimable work.

pub fn ready_items(items: Vec<WorkItem>, now: DateTime<Utc>) -> Vec<WorkItem>

pub fn validate_link( edge: &WorkEdge, existing_items: &[WorkItem], existing_edges: &[WorkEdge], ) -> Result<(), WorkGraphError>

Trait Implementations

impl Clone for WorkGraphMachine

fn clone(&self) -> WorkGraphMachine

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Copy for WorkGraphMachine

impl Debug for WorkGraphMachine

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

Formats the value using the given formatter.

impl Default for WorkGraphMachine

fn default() -> WorkGraphMachine

Returns the “default value” for a type.