These are chat archives for akkadotnet/akka.net

31st
Aug 2018
Vagif Abilov
@object
Aug 31 2018 10:58
@Aaronontheweb "so System.String isn't in the same place on .NET Framework and .NET Core that's what that serializer accounts for"
But is it System.String on all platforms, even though it resides in different assemblies. Shouldn't this be sufficient for a serializer?
atresnjo
@atresnjo
Aug 31 2018 14:01
is there a way to know if an actor is recovering from a journal? I've found the IsRecovering property, but it's useless for me because it's set to true when creating an actor.
Pieter Joost van de Sande
@pjvds
Aug 31 2018 15:36
any examples of how to use Become with the F# api?
nathvi
@nathvi
Aug 31 2018 15:49
Hello all
Bartosz Sypytkowski
@Horusiath
Aug 31 2018 16:08
@pjvds F# has mutually recursive functions and pattern matching - so you don't really need Become, you can just use F# patterns for working with state machines.
@atresnjo what's wrong with that?
Pieter Joost van de Sande
@pjvds
Aug 31 2018 16:15
@Horusiath you mean like:
let a (mailbox : Actor<_>) =
        let rec loop () =
            actor {
                let! ctx = mailbox.Receive()
                "hello from A" |> mailbox.Sender().Tell
                return! b mailbox
            }
        loop(count)

let b (mailbox : Actor<_>) =
        let rec loop () =
            actor {
                let! ctx = mailbox.Receive()
                "hello from B" |> mailbox.Sender().Tell
                return! a mailbox
            }
        loop(count)
atresnjo
@atresnjo
Aug 31 2018 17:12
@Horusiath Because when an actor is created I save a snapshot of the state; but I want to avoid saving a snapshot of it when it gets created through the recovery process.
Robert Morris
@BobbyMorris
Aug 31 2018 19:41
we're using akka.di.castlewindsor and we've got leaks, i've been able to mitigate them by having actors turn refs of themselves into a static container ref for release during actor poststop. ants shows the deps we have hanging around and they all should be registered as transient and they all have idisposables in their graph, and they all are referenced by the container kernel.
barring other problems shouldn't i be expecting the actor system to release actors to the container when stopped already?
if i have to turn everything in that was resolved myself does putting all cleanup in poststop cover it?