Runic.Workflow.Runnable (Runic v0.1.0-alpha.7)

A prepared unit of work ready for execution.

Contains everything needed to execute independently of the source workflow. After execute/2, contains result and events for reducing back into workflow.

Three-Phase Execution Model

Prepare - Extract minimal context from workflow, build a Runnable
Execute - Run the node's work function in isolation (potentially parallel)
Apply - Fold events back into the workflow via apply_event/2

The Runnable struct is the carrier between these phases, holding:

The node to invoke
The input fact triggering invocation
Minimal causal context (no full workflow reference)
After execution: result, status, and events for reducing into workflow

Summary

Types

status()

t()

Functions

complete(runnable, result, events)

Marks a runnable as completed with result and events.

complete(runnable, result, events, hook_apply_fns)

Marks a runnable as completed with events and hook apply_fns.

fail(runnable, error)

Marks a runnable as failed with an error.

new(node, fact, context)

Creates a new Runnable in pending state.

new(id, node, fact, context)

Creates a new Runnable with explicit id.

runnable_id(node, fact)

Generates a stable runnable id from node and fact hashes.

skip(runnable, events)

Marks a runnable as skipped with events.

Types

status()

@type status() :: :pending | :completed | :failed | :skipped

t()

@type t() :: %Runic.Workflow.Runnable{
  context: Runic.Workflow.CausalContext.t() | nil,
  error: term() | nil,
  events: [struct()] | nil,
  hook_apply_fns: [function()] | nil,
  id: integer() | nil,
  input_fact: Runic.Workflow.Fact.t(),
  node: struct(),
  result: term() | nil,
  status: status()
}

Functions

complete(runnable, result, events)

@spec complete(t(), term(), [struct()]) :: t()

Marks a runnable as completed with result and events.

Events are the list of event structs produced by Invokable.execute/2. They will be folded into the workflow via apply_event/2 during the apply phase.

complete(runnable, result, events, hook_apply_fns)

@spec complete(t(), term(), [struct()], [function()]) :: t()

Marks a runnable as completed with events and hook apply_fns.

fail(runnable, error)

@spec fail(t(), term()) :: t()

Marks a runnable as failed with an error.

new(node, fact, context)

@spec new(struct(), Runic.Workflow.Fact.t(), Runic.Workflow.CausalContext.t()) :: t()

Creates a new Runnable in pending state.

The id is a hash of {node.hash, fact.hash} for idempotency tracking.

new(id, node, fact, context)

@spec new(
  integer(),
  struct(),
  Runic.Workflow.Fact.t(),
  Runic.Workflow.CausalContext.t()
) :: t()

Creates a new Runnable with explicit id.

runnable_id(node, fact)

@spec runnable_id(
  struct(),
  Runic.Workflow.Fact.t()
) :: integer()

Generates a stable runnable id from node and fact hashes.

skip(runnable, events)

@spec skip(t(), [struct()]) :: t()

Marks a runnable as skipped with events.

The events (typically just ActivationConsumed) are folded during apply, and downstream nodes are marked as :upstream_failed.