pekko-py: Pekko Typed style actors for Python asyncio

[hanishi]

https://github.com/hanishi/pekko-py

With Python's match, frozen dataclasses, and type unions, it reads close to the Scala DSL:

@dataclass(frozen=True)
class Increment:
    by: int

@dataclass(frozen=True)
class GetTotal:
    reply_to: ActorRef[int]

type Command = Increment | GetTotal

def counter(total: int = 0) -> Behavior[Command]:
    def on_message(msg: Command) -> Behavior[Command]:
        match msg:
            case Increment(by):
                return counter(total + by)
            case GetTotal(reply_to):
                reply_to.tell(total)
                return Behaviors.same
    return Behaviors.receive_message(on_message)

[pjfanning]

Looks interesting. Could you update your README to say Apache Pekko when referring to our tools? Would you be able to add a disclaimer to your README saying that your repo is not affiliated with Apache Pekko, just to avoid confusion?

[pjfanning]

Looks like https://pypi.org/project/tapio-py/ has also been created recently

[hanishi]

Looks interesting. Could you update your README to say Apache Pekko when referring to our tools? Would you be able to add a disclaimer to your README saying that your repo is not affiliated with Apache Pekko, just to avoid confusion?

Thanks for pointing that out! I already included a disclaimer at the bottom of the README, but I’ll make it more prominent by moving it to the top in bold. I’ll also make sure to refer to Apache Pekko by its full name.

Looks like https://pypi.org/project/tapio-py/ has also been created recently

Another interesting direction would be MicroPython support. Imagine lightweight actors running on an ESP32, communicating with actors on the JVM through Apache Pekko. That could open up some interesting use cases for IoT.

[hanishi]

🚀 pekko-py now supports clustering!

What started as an experiment to bring the Typed Actor model to Python has evolved into a distributed Actor runtime.

Cluster features now include:

• TCP Remoting and remote ActorRefs
• Cluster Membership and Gossip Protocol
• Vector Clocks and Phi Accrual Failure Detection
• Cluster Singleton
• Cluster Sharding with rebalancing and handoff
• Split Brain Resolver (KeepMajority, StaticQuorum, etc.)
• Cluster Receptionist
• Kubernetes Discovery and Cluster Bootstrap
• Coordinated Shutdown

The API follows Apache Pekko Typed concepts while taking advantage of Python’s native asyncio ecosystem.

I’m interested in exploring how the Actor model can complement Python’s numerical computing ecosystem, including NumPy and distributed stateful workloads.

Feedback from people familiar with Pekko internals and distributed systems would be especially welcome!

🔗 https://github.com/hanishi/pekko-py

[hanishi]

It's secured.
https://pypi.org/project/ohra/