Where communities thrive


  • Join over 1.5M+ people
  • Join over 100K+ communities
  • Free without limits
  • Create your own community
People
Activity
  • 20:57
    IgorFedchenko commented #4085
  • 20:32
    IgorFedchenko commented #4085
  • 20:01
    IgorFedchenko commented #4085
  • 19:55
    IgorFedchenko commented #4085
  • 16:22
    Aaronontheweb labeled #3997
  • 16:22
    Aaronontheweb closed #3997
  • 16:20
    IgorFedchenko commented #3997
  • 16:08
    IgorFedchenko commented #4085
  • 15:50
    Aaronontheweb assigned #4085
  • 15:50
    Aaronontheweb labeled #4085
  • 15:50
    Aaronontheweb labeled #4085
  • 15:49
    Aaronontheweb closed #4032
  • 14:58
    IgorFedchenko commented #4032
  • 14:57
    IgorFedchenko opened #4085
  • Dec 05 17:21
    Aaronontheweb synchronize #4079
  • Dec 05 17:20
    Aaronontheweb labeled #4084
  • Dec 05 17:20
    Aaronontheweb labeled #4084
  • Dec 05 17:20
    Aaronontheweb milestoned #4084
  • Dec 05 17:20

    Aaronontheweb on dev

    Remove string interpolation fro… (compare)

  • Dec 05 17:20
    Aaronontheweb closed #4084
jcwrequests
@jcwrequests
Currently I am working on a patch for the current implementation which should be available soon.
jcwrequests
@jcwrequests
@tomstaijen That was not a very good explanation and I apologize. I am working on a Patch as I previously said for DI and within that I am adding some examples. So hopefully by tomorrow evening EST I should have something for you to look at.
Tom Staijen
@tomstaijen
@jcwrequests Thanks, it helped, found the resolver in de context's props. Eagerly waiting to see your patch. W.r.t. to lifetime, for now i'm working with singleton services and forced per-message lifetimescope (using Autofac's Func<Owned<T>>)
Roger Johansson
@rogeralsing
akkagraph.png
Aaron Stannard
@Aaronontheweb
@rogeralsing that's super cool! How'd you make that?
Roger Johansson
@rogeralsing
here is a better one
akkagraph.png
played around with a "Walk" system message, that notifies a callback of its existence, and then do the same for its own children.. then I rendered the result in GraphViz
but it could be solve with proper monitor support instead, that would be better. if we have a IMonitorLifecycle interface that gets called from the cell whenever a lifecycle event occurs, the one could plug that in there and get a realtime view :)
Roger Johansson
@rogeralsing
We should totally provide a debugger like this, like a vs debug extension. hover a system or actorref instance and get a full view of the system :)
Aaron Stannard
@Aaronontheweb
yeah, I think that's a really cool idea
I took a look at @jcwrequests post on monitoring via AOP the other day
that's an idea I think we should explore more closely after 1.0 ships
Bartosz Sypytkowski
@Horusiath
considering #722 - I've created a test case for app.config HOCON string loading... but it passes anyway, config is loaded correctly
Aaron Stannard
@Aaronontheweb
@/all for anyone interested in @jcwrequests's post about AOP for instrumenting monitoring in Akka.NET: http://www.message-flow.com/2015/01/monitoring-akkanet-based-systems.html
@Horusiath hmmm... can @jorgef replicate the issue?
because App.config loading is only automatic in /dev right now
Akka.NET v0.8 on NuGet doesn't do it
Bartosz Sypytkowski
@Horusiath
I'm testing explicit loading with F# Configuration.load() which is basically an alias for ConfigurationFactory.Load()
also, when @HCanber schedulers can be merged to dev branch?
Aaron Stannard
@Aaronontheweb
@Horusiath as soon as we figure out why @HCanber's pull request makes the system load performance counter throw an InvalidOperationException then I think we can move on
and merge it in
which is really weird, because his code doesn't even touch the performance counter definitions
at least in a way that's obvious
Bartosz Sypytkowski
@Horusiath
fu*k, I have to include JSON.NET converter for F# Expr but the problem is... Akka.FSharp has no direct dependency to JSON.NET :D
Roger Johansson
@rogeralsing
@Horusiath why do you need that? do you need to do some special handling with the F# expressions?
Bartosz Sypytkowski
@Horusiath
expression-based deciders i.e.
Roger Johansson
@rogeralsing
but isnt the akka serializer mechanics enough to deal with that? do you need to do actual json parsing/writing/annotations?
Bartosz Sypytkowski
@Horusiath
but this is so minor feature, I think I'll drop this idea untill there will be something more justifying reason to include Expr support for JSON.NET
NetwonSoftJsonSerializer will not know how to serialize F# quotation expressions
only FsPickler does
Roger Johansson
@rogeralsing
oh, and we removed fspickler?
Bartosz Sypytkowski
@Horusiath
no, but it used on top-level only, as attached akka serializer
expressions being properties of serialized objects won't work with it
Roger Johansson
@rogeralsing
ah, that should probably be possible to solve using the surrogate mechanism now. eg. FSExpressionDecider (containing expressions) -> FSExpressionDeciderSurrogate containing expressions pre serialized by fs pickler and stored as a string base64 or whatever in the surrogate., then the other way around, turn the base64 string into expressions and recreate the FSExpressionDecider
Bartosz Sypytkowski
@Horusiath
yes, but still it will work only for special case (Deciders), while leaving potential hole for other types of messages using expressions
jcwrequests
@jcwrequests
@rogeralsing That is and awesome way to do diagrams and map system flow and that is definitely more useful the any UML diagram.
jcwrequests
@jcwrequests
@Aaronontheweb @rogeralsing @Horusiath @HCanber @stefansedich The DI patch is working even with the AutoFac. Currently I am putting together some examples using the containers. Currently I have implement examples with HashPool and one with Child Actors for @tomstaijen. I want to add a few more examples and clean things up a bit before putting in a proper PR. @tomstaijen If you are interested or can't wait for the PR you can check out the latest here
Roger Johansson
@rogeralsing
/all are we really handling supervision correctly when it comes to restarting children? e.g. lets say we have the structure a -> b -> c where c is stashable. and actor c throws, b supervises but escallate to a.. A decides to restart b, b will now recreate c and thus we loose any stashing as it is no longer the same c.. is this the behavior in jvm akka too?
Bartosz Sypytkowski
@Horusiath
I think we should remove scheduling functions from F# API, after new schedulers API they have basically no value added
Bartosz Sypytkowski
@Horusiath
also @HCanber what is the relation between Scheduler and Scheduler.Advanced? Some tips, when which of them should be used?
Aaron Stannard
@Aaronontheweb
@rogeralsing you mean is it the right behavior for an actor to kill off its children during restarts?
jcwrequests
@jcwrequests
@Aaronontheweb Do you think I should add more examples or do you think this is enough? Any Advice before I put in my PR. Thanks
Aaron Stannard
@Aaronontheweb
I think that's probably good @jcwrequests
Roger Johansson
@rogeralsing
@aaronontheweb, yes. It is always only the direct children of the decidung supervisor that restarts, grand* children are completely replaced
Not saying this is wrong, just checking if that is the expected behavior. It could be suprising if you have stashing and the actor throws and you loose all messages because the exception was escalated
Aaron Stannard
@Aaronontheweb
@rogeralsing I've thought that this is weird behavior too to be honest
Håkan Canberger
@HCanber
@Horusiath The operations directly under Scheduler only contains scheduling Tells, while Advanced contains scheduling actions.
Roger Johansson
@rogeralsing
Is there any way to show the build bot result in some other way? it is super duper spammy in the github notification log.. 75% buildbot events, 25% human interaction
spam.png