Implement by handling incoming commands and return an Effect() to persist or signal other effects of the command handling such as stopping the behavior or others.

Use EventSourcedBehavior#newCommandHandlerBuilder to define the command handlers.

The command handlers are only invoked when the actor is running (i.e. not replaying). While the actor is persisting events, the incoming messages are stashed and only delivered to the handler once persisting them has completed.

Implement by returning the initial empty state object. This object will be passed into this behaviors handlers, until a new state replaces it.

Also known as "zero state" or "neutral state".

Implement by applying the event to the current state in order to return a new state.

Use EventSourcedBehavior#newEventHandlerBuilder to define the event handlers.

The event handlers are invoked during recovery as well as running operation of this behavior, in order to keep updating the state state.

For that reason it is strongly discouraged to perform side-effects in this handler; Side effects should be executed in thenRun or recoveryCompleted blocks.

Factory of effects.

Return effects from your handlers in order to instruct persistence on how to act on the incoming message (i.e. persist events).

INTERNAL API: DeferredBehavior init

Transform the event in another type before giving to the journal. Can be used to wrap events in types Journals understand but is of a different type than Event.

Narrow the type of this Behavior, which is always a safe operation. This method is necessary to implement the contravariant nature of Behavior (which cannot be expressed directly due to type inference problems).

Override to change the strategy for recovery of snapshots and events. By default, snapshots and events are recovered.

Criteria for retention/deletion of snapshots and events. By default, retention is disabled and snapshots are not saved and deleted automatically.

Initiates a snapshot if the given predicate evaluates to true.

Decide to store a snapshot based on the State, Event and sequenceNr when the event has been successfully persisted.

When persisting multiple events at once the snapshot is triggered after all the events have been persisted.

Snapshots triggered by snapshotWhen will not trigger deletes of old snapshots and events if EventSourcedBehavior.retentionCriteria with RetentionCriteria.snapshotEvery is used together with shouldSnapshot. Such deletes are only triggered by snapshots matching the numberOfEvents in the RetentionCriteria.

Override to react on general lifecycle signals and persistence specific signals (subtypes of akka.persistence.typed.EventSourcedSignal).

Use EventSourcedBehavior#newSignalHandlerBuilder to define the signal handler.

Transform the state into another type before giving it to and from the journal. Can be used to migrate from different state types e.g. when migration from PersistentFSM to Typed EventSourcedBehavior.

Transform the incoming messages by placing a funnel in front of the wrapped Behavior: the supplied PartialFunction decides which message to pull in (those that it is defined at) and may transform the incoming message to place them into the wrapped Behavior’s type hierarchy. Signals are not transformed.

Example:

val b: Behavior[Number] =
  Behaviors
    .receive[String] { (ctx, msg) =>
      println(msg)
      Behaviors.same
    }
    .transformMessages[Number] {
      case b: BigDecimal => s"BigDecimal($b)"
      case i: BigInt     => s"BigInteger($i)"
      // all other kinds of Number will be `unhandled`
    }

The ClassTag for Outer ensures that only messages of this class or a subclass thereof will be intercepted. Other message types (e.g. a private protocol) will bypass the interceptor and be continue to the inner behavior untouched.

Override and define the snapshot selection criteria used by this actor instead of the default. By default the most recent snapshot is used, and the remaining state updates are recovered by replaying events from the sequence number up until which the snapshot reached.

You may configure the behavior to skip replaying snapshots completely, in which case the recovery will be performed by replaying all events -- which may take a long time.