I think the gateway would be way more useful (and potentially more private) if instead of provoding WS for replicating a hyperdrive, it provided a WS interface to discovery-swarm.
I'm thinking this should be built as a separate module first.
It shouldn't reveal peer IP addresses
It should work over a stream.
All peer connections should be multiplexed over the single stream
Have the constructor take a discovery-swarm instance and return a handleStream(stream) listener. (or something like that)
It should use protocol buffers for an RPC API
join(discoveryKey) -> To the gateway
leave(discoveryKey) -> To the gateway
peer(discoveryKey, peerId) -> From the gateway
connect(peerId, streamId) -> To the gateway
disconnect(peerId, streamId) -> To the gateway
data(peerId, streamId, data) -> Either way
The client-side should take a stream as an argument and expose the same API as discovery-swarm
I think the gateway would be way more useful (and potentially more private) if instead of provoding WS for replicating a hyperdrive, it provided a WS interface to discovery-swarm.
I'm thinking this should be built as a separate module first.
discovery-swarminstance and return ahandleStream(stream)listener. (or something like that)The client-side should take a stream as an argument and expose the same API as
discovery-swarm