Why does distributed.rs appear to send two messages when replying to a remote actor?

[phil-skillwon]

On line 116:
Self::Msg::Rpc(request, reply) => {
tracing::info!(
"Received an RPC of '{request}' replying in kind to {} remote actors",
remote_actors.len()
);
let reply_msg = format!("{request}.");
reply.send(reply_msg.clone())?;
for act in remote_actors {
let _reply =
ractor::call_t!(act, PingPongActorMessage::Rpc, 100, reply_msg.clone())?;
}
}
at first glance, it seems like a message is sent once: reply.send,
and then a similar message is sent again to all remote actors: ractor::call_t!(act, PingPongActorMessage::Rpc, 100, reply_msg.clone())?;
I'm a bit confused. Apart from replying in this manner, are there any other methods that can be used as an alternative?

[slawlor]

RPC's must have a reply, while non-rpc's don't.

So we first reply, then that actor triggers a broadcast to the other actors waiting on their replies. It's just using RPCs instead of cast operations. It's not really probably the best example, but it's a playground, it's not meant to be production examples.