# ChangeEventHandler no-op

**URL:** <https://discuss.akka.io/t/changeeventhandler-no-op/10620>\
**Category:** Akka Libraries\
**Created:** [February 9, 2024, 6:48pm UTC](https://discuss.akka.io/t/changeeventhandler-no-op/10620 "2024-02-09T18:48:02Z")\
**Posts on this page:** 6\
**Page:** 1

<div class="post-metadata">

**Author:** ![dleacock](https://sea2.discourse-cdn.com/flex020/user_avatar/discuss.akka.io/dleacock/32/2570_2.png) [@dleacock](https://discuss.akka.io/u/dleacock)\
**Post date:** [February 9, 2024, 6:48pm UTC](https://discuss.akka.io/t/changeeventhandler-no-op/10620/1 "2024-02-09T18:48:02Z")

</div>

I’m currently building a small project to understand how to best use an Akka Durable State entity alongside with a `ChangeEventHandler` so I can project journal events downstream. I wanted to add query Command that would return the latest State information back to the caller. However when I do that I am forced to handle that case in the `ChangeEventHandler` and persist an event for this query. That event wouldn’t really be an event about the system and would just clog up the journal.

```auto
trait Command
case class UpdateEntity(id: UUID, payload: String) extends Command
case class GetEntity(id: UUID): String extends Command

trait Event
case class EntityUpdated(payload: String) extends Event
case class EntityStateRetrieved extends Event // This seems like a wasteful event to persist in the journal

private val stateChangeEventHandler = ChangeEventHandler[Command, State, Event](
   updateHandler = {
          case (_, _, UpdateEntity(id, payload)) => EntityUpdated(payload) 

          case (_, _, GetEntity(id)) => EntityStateRetrieved //I'm forced to handle this and persist an event for a GET command,
   },
   deleteHanlder = { ... }
)

```

What is the best way to approach this? I would prefer to get the state through the entity itself but I don’t want to be forced to persist an event that doesn’t match the domain behaviour of the entity. I would have thought that it may be similar to event-sourced behaviour `Effect.persist(Seq[Event])` where Seq could be empty.

Thank you

---

<div class="post-metadata">

**Author:** ![leviramsey](https://sea2.discourse-cdn.com/flex020/user_avatar/discuss.akka.io/leviramsey/32/2653_2.png) [@leviramsey](https://discuss.akka.io/u/leviramsey)\
**Post date:** [February 9, 2024, 9:42pm UTC](https://discuss.akka.io/t/changeeventhandler-no-op/10620/2 "2024-02-09T21:42:34Z")

</div>

After looking through the implementation, the `updateHandler` should only be called if there’s an `Effect.persist` (vs., for instance, `Effect.none`) returned from the command handler.

So I think is just the Scala compiler warning about exhaustiveness of the pattern match, in which case you could satisfy the compiler with something like:

```scala
case (_, _, GetEntity(_)) => ??? // GetEntity does not update state, so this should never be hit

```

(or if some linting about `???` is in use, an equivalent `throw`: `AssertionError` (to indicate that this (`GetEntity` updating state) is a bug in the code) might be the most appropriate throwable for this).

Conceivably the implementation could change to compare the new state with the old state and only call the `updateHandler` if there’s a change, though that depends on `equals` being implemented (and there’s some recent controversy in the DDD world about whether `equals`/`hashCode` should be overridden for entities…).

---

<div class="post-metadata">

**Author:** ![dleacock](https://sea2.discourse-cdn.com/flex020/user_avatar/discuss.akka.io/dleacock/32/2570_2.png) [@dleacock](https://discuss.akka.io/u/dleacock)\
**Post date:** [February 13, 2024, 1:18am UTC](https://discuss.akka.io/t/changeeventhandler-no-op/10620/3 "2024-02-13T01:18:54Z")

</div>

That makes sense, thank you for the sanity check. I may have also just done something wrong in my implementation and misunderstood the error I was getting.

---

<div class="post-metadata">

**Author:** ![johanandren](https://sea2.discourse-cdn.com/flex020/user_avatar/discuss.akka.io/johanandren/32/85_2.png) [@johanandren](https://discuss.akka.io/u/johanandren)\
**Post date:** [February 19, 2024, 7:53am UTC](https://discuss.akka.io/t/changeeventhandler-no-op/10620/4 "2024-02-19T07:53:25Z")

</div>

I think that it is an oversight in the new `ChangeEventHandler` API that you cannot signal that there is no change for a command, I’ve created an issue to track improving this: [No way to signal no change with ChangeEventHandler · Issue #32310 · akka/akka · GitHub](https://github.com/akka/akka/issues/32310)

---

<div class="post-metadata">

**Author:** ![dkichler](https://sea2.discourse-cdn.com/flex020/user_avatar/discuss.akka.io/dkichler/32/1808_2.png) [@dkichler](https://discuss.akka.io/u/dkichler)\
**Post date:** [February 20, 2024, 4:24pm UTC](https://discuss.akka.io/t/changeeventhandler-no-op/10620/5 "2024-02-20T16:24:39Z")

</div>

I might suggest it should mimic the API of `Effect.persist`, accepting a `Seq[Event]`, allowing the client to persist 0 to many events.

---

<div class="post-metadata">

**Author:** ![johanandren](https://sea2.discourse-cdn.com/flex020/user_avatar/discuss.akka.io/johanandren/32/85_2.png) [@johanandren](https://discuss.akka.io/u/johanandren)\
**Post date:** [February 20, 2024, 4:53pm UTC](https://discuss.akka.io/t/changeeventhandler-no-op/10620/6 "2024-02-20T16:53:37Z")

</div>

As Patrik mentioned in the issue, that is not possible as change events sequence numbers must be in sync with the durable state revision number. It must always be a 1:1 with actual updates.

I don’t think there is a way to design our way out of that, so we will document more clearly that commands not leading to an update is never fed to the `ChangeEventHandler` but that the type system can’t know that and catch-throw-illegal-argument or something is reasonable.