daffi

daffi is a high-performance Python RPC framework with a native compiled core.
It lets processes call each other's functions over TCP or Unix sockets — with minimal boilerplate and near-zero overhead.

---

Key features

Feature	Details
Native core	The message framing, connection management and routing logic is implemented in a native compiled extension (dfcore) for maximum throughput.
Multiple transports	TCP, Unix domain sockets, TLS — same API.
Four call styles	rpc, rpc_nowait, cast, cast_nowait — blocking/non-blocking, single target/broadcast.
Four serialisation formats	PICKLE (Python-only, default), JSON, MSGPACK, OPAQUE (raw bytes or string).
Event subscriptions	React to nodes joining and leaving the network.

---

Install

pip install daffi

# optional: MessagePack support
pip install "daffi[msgpack]"

---

30-second example

Service — expose a function:

from daffi import Service, callback

@callback
def add(a: int, b: int) -> int:
    return a + b

svc = Service(app_name="calc", host="127.0.0.1", port=5000)
svc.start()
svc.join()

Client — call it:

from daffi import Client

client = Client(app_name="my-client", host="127.0.0.1", port=5000)
conn = client.connect()

result = conn.rpc(timeout=5).add(3, 4)   # → 7
client.stop()

---

Two topologies

daffi supports two network topologies:

• Client → Service — the Service hosts callbacks; Clients call them directly.
  Best for simple request/response scenarios.

• Client → Router → Worker — a Router forwards calls between Callers and Workers.
  Workers can be added or removed at runtime without restarting anything.
  Best for scalable, distributed workloads.

See Architecture for diagrams and a deeper explanation.