Struct protosocket_server::Server

pub struct Server { /* private fields */ }

Implementations

impl Server

pub fn new() -> Result<Self>

pub fn register_service_listener<Connector: ServerConnector>( &mut self, listener: TcpListener, server_state: Connector, ) -> Result<ConnectionServer<Connector>>

listener must be configured non_blocking

see register_multithreaded_service_listener for a configuration that has multiple IO drivers for the service.

pub fn register_multithreaded_service_listener<Connector: ServerConnector + Clone>( &mut self, listener: TcpListener, server_state: Connector, thread_count: usize, ) -> Result<Vec<ConnectionServer<Connector>>>

Configure a single tcp listener to vend connections to separate IO drivers (one driver per connection server).

Each of the returned ConnectionServers can be spawned onto distinct threads, and their connections will be segregated. Connections are vended to the backing threads in round-robin fashion. There is no work stealing between the ConnectionServers - they are fully distinct.

ConnectionServers can reuse the same task work pool if desired, or they can use separate async work pools. What works best for you is what you should do.

listener must be configured non_blocking

pub fn serve(self) -> Result<()>

Dedicate a thread to your connection frontend.