Where communities thrive


  • Join over 1.5M+ people
  • Join over 100K+ communities
  • Free without limits
  • Create your own community
People
Activity
  • Dec 17 2020 21:21
    andreaTP commented #126
  • Dec 17 2020 19:39
    vladimir-ivanov commented #126
  • Dec 17 2020 19:38
    vladimir-ivanov commented #126
  • Dec 17 2020 19:34
    vladimir-ivanov closed #126
  • Dec 17 2020 19:34
    vladimir-ivanov commented #126
  • Dec 17 2020 19:32
    andreaTP commented #126
  • Dec 17 2020 19:18
    vladimir-ivanov opened #126
  • Oct 15 2020 13:57
    tometchy commented #125
  • Oct 15 2020 13:56
    tometchy commented #125
  • Oct 15 2020 13:44
    andreaTP commented #12
  • Oct 15 2020 13:01
    bulkazszynka commented #12
  • Oct 15 2020 12:58
    andreaTP commented #125
  • Oct 15 2020 03:34
    tometchy closed #125
  • Oct 15 2020 03:34
    tometchy commented #125
  • Oct 14 2020 22:58
    andreaTP commented #125
  • Oct 14 2020 12:31
    tometchy opened #125
  • Sep 30 2020 12:30

    andreaTP on v2.2.6.9

    (compare)

  • Sep 30 2020 12:29

    andreaTP on master

    Bump: Update to upstream Akka 2… (compare)

  • Sep 30 2020 12:29
    andreaTP closed #124
  • Sep 29 2020 13:25
    andreaTP opened #124
Marco F.
@mfirry
i guess i figured fastparse 1 is long gone
which is half true
scalameta maintains a fastparse 1 fork, as far as i understood
Marco F.
@mfirry
what do you think @FabioPinheiro ?
Fabio Pinheiro
@FabioPinheiro

I assumed (the migration to fastparse2 and 2.13) was in the same branch
But Yes is better to have two commits. I will make a PR to support Scala 2.13 today based in your branch @mfirry

@andreaTP Will you created a new main branch in addition to master where the support to Scala native is drooped? or should we make the PR to master?

Andrea Peruffo
@andreaTP
let's work in master, we can re-add SN later on if needed/requested
Marco F.
@mfirry
hang on hang on. i'm lost

I assumed (the migration to fastparse2 and 2.13) was in the same branch

it is. i mean same branch but i wanted fastparse2 to work BEFORE i start changing scala version

let me do something
I'll merge the branch on my master
and start the WIP PR
Andrea Peruffo
@andreaTP
Akka.Js 1.2.6.1 is released :-)
and support for Scala 2.13 is ready (cc. @FabioPinheiro )but blocked by: wvlet/airframe#868
Andreas Gabor
@an-tex
hey @andreaTP ! I'm getting this when not having the snapshot resolver added with the latest 1.2.6.1 [error] not found: .ivy2/local/org.akka-js/shocon_sjs0.6_2.12/0.5.1-SNAPSHOT/ivys/ivy.xml
[error] not found: https://repo1.maven.org/maven2/org/akka-js/shocon_sjs0.6_2.12/0.5.1-SNAPSHOT/shocon_sjs0.6_2.12-0.5.1-SNAPSHOT.pom
Andrea Peruffo
@andreaTP
argh ...
right
I published without updating shocon :-(
try to add: "org.akka-js" %%% "shocon" % "0.5.0"
cc. @an-tex
Andrea Peruffo
@andreaTP
publishing 2.2.6.1 as a fix (and a major bump is good here)
2.2.6.1 published on sonatype
Andreas Gabor
@an-tex
cheers mate! works
I guess it's on your list but just as I stumbled over it https://github.com/akka-js/akka.js/releases is missing the latest release
Andrea Peruffo
@andreaTP
I missed the tag, thanks @an-tex everything should be fine now
Volkan Unsal
@volkanunsal
Hello. I have a couple of quick question regarding identifying actors by id. Given I have an actor system that runs on several worker threads, can I identify each actor uniquely by some id or a url, e.g. node address + actor id?
Second, I'm going to be running this app on a Kubernetes (AWS Fargate) cluster, which may scale up and down depending on resource utilization. How would the Actor system be affected by this activity? Are there any guideline for setting up an actor system on Kubernetes?
Andrea Peruffo
@andreaTP
Separate worker threads are separate processes in Node/browser, there is no way to identify by id an actor running in another VM unless we set-up something like an akka-cluster
Kubernetes is outside the scope of this project, this is a toolkit to develop applications, and since that akka-cluster is not available how the application handle restarts is all in the logic that you are going to write
Volkan Unsal
@volkanunsal
I see. Thanks for clarifying that. So if I wanted to use clustering, I'd need to use Akka on JVM, correct?
Andrea Peruffo
@andreaTP
Correct
And I encourage you to use the officially supported Akka release for production systems
Volkan Unsal
@volkanunsal
Thank you. I think I might "hack" a cluster by using the Node hostname as the name of the actor system and use an external in-memory database server as a source of truth. I don't need actor to actor communication, just need to track actors across the node workers.
Andrea Peruffo
@andreaTP
You can do that, of course, but Akka cluster exists and Akka management make it super easy to deploy to Kubernetes ...
There is no real need to hack things around if you stay on the JVM
Also if you wanna write business logic in JS (I'm assuming that's the real reason for those questions) GraalVM provides a polyglot JVM that might help
Do you wanna elaborate on the use case @volkanunsal ?
Volkan Unsal
@volkanunsal
I have a Node.js codebase that uses worker threads for concurrency, which don't scale well, and I want to use a more lightweight concurrency model. The actors would be used to create a Cometd subscription to a long-polling server for every object inside an S3 bucket. Each actor will be responsible for holding a single subscription, and when a new replica is launched, I want to make sure that the actor system knows which objects are "taken" by an actor, and avoid subscribing that object again.
So my hack is to use this.path() of the actor, and store that in a Redis backend with the object id as a central database of where each object subscription is located. And I plan to use that in a very limited way, i.e. only as a blacklist, for now.
In one of your talks, you mentioned that we can configure the actor system to use as many worker threads as we like. Is there an example of how to do that?
Andrea Peruffo
@andreaTP
yes, here: https://github.com/akka-js/akka.js-dom
but is focused on the browser side
looks like a pretty heavy production system, I suggest you to switch to something more robust and well maintained like Akka for the JVM, but, well if you continue with Akka.Js, well keep pinging me and I’ll try to do my best to support :-)
Volkan Unsal
@volkanunsal
Very kind of you. I'll keep you posted if I run into any issues. btw, I think this project is pretty awesome.
Quck question. Is there any support for supervisor strategy?
Andrea Peruffo
@andreaTP
Supervisor strategy is all there, though it doesn’t have a cleaned up API
(from JS I mean) if you use Scala is all there
it should be just a matter of exposing the API through akka.js bindings @volkanunsal
Volkan Unsal
@volkanunsal
Is it possible to have multiple actor systems working at the same time? In my tests, I can only have one active system at a time.
Volkan Unsal
@volkanunsal
nvm, I think it was a fluke.
Andrea Peruffo
@andreaTP
multiple ActorSystems are perfectly fine
Volkan Unsal
@volkanunsal
quick question. I'm trying to write tests for the actors. Since they're launched in a separate worker thread, I'm having difficulties with it. How can I tell the actor system to launch the workers in the current thread?
Andrea Peruffo
@andreaTP
Testkit is pretty limited ATM, basically it’s taking control over the thread of the JS VM, so, you can’t really do nothing with the “real world” since the external calls are not handled.
My suggestion would be to separate the tests:
  • use testkit exclusively to test actor’s business logic without interactions with the external world
  • use other tests to test exclusively the interactions