How much of the expected feature set is completed?

[seivan]

First of, amazing work on the API design, you clearly understand the assignment on what's missing for Rust.

I believe this should be the foundation for most of the backend frameworks from now on, it would give the same ecosystem benefits that Elixir had for Phoenix.

But I have a few questions... coming from Elixir, how many of the features are functional and stable now?

1. Does gossiping for node connections work?
2. Is global available for cluster wide registration of GenServers?
3. Can Server A send a message to a GenServer created on Server B to mutate state, and expect a response?
4. Can Server A monitor a GenServer on Server B to relaunch if it fails?
5. Can I use k8 headless services or ECS DNS lookups to connect Node 6 to Node 5, and get Nodes 1-4 as well?

Just trying to wrap my head around what features are missing and how stable the networking features are.

In fact most of my concerns are the networking stuff, if the distributed processes have any issues.

[dtzxporter]

1. Yes, though no connection is made until an attempt to send to the discovered node (or monitor, etc) eventually will be configurable like erlang.
2. No, registries are available at the moment and you can use named processes to talk to them remotely.
3. Yes, works fully.
4. It can monitor it, but you would need some logic to restart it, you can't 'spawn' random code remotely of course because rust isn't dynamic.
5. You can manually manage nodes using the Node methods: https://docs.rs/hydra/latest/hydra/enum.Node.html#method.connect