Struct C2ReconnectingHandle 

pub struct C2ReconnectingHandle { /* private fields */ }

Swappable, never-closing counterpart to C2ControlHandle. Every accessor clones or reads out of a watch cell a background supervisor mutates on reconnect; no caller ever holds a handle bound to one specific physical connection, so a relay restart never permanently breaks it.

Implementations

impl C2ReconnectingHandle

pub fn link_state(&self) -> C2LinkState

Whether the underlying physical connection is live right now.

pub fn link_state_receiver(&self) -> Receiver<C2LinkState>

A fresh receiver over the same link-state cell, for a caller that wants to await transitions rather than poll.

pub fn current_topology(&self) -> Arc<C2Topology>

pub fn subscribe_topology(&self) -> Receiver<Arc<C2Topology>>

pub async fn browser_command( &self, cmd: BrowserClientCommand, ) -> Result<(), C2ControlError>

Validates and synchronously enqueues one typed request against whichever physical connection is live right now. Returns Err(C2ControlError::Closed) immediately – never blocks or waits for a reconnect – while the link is down.

pub fn start_request( &self, route: NodeRoute, request: NodeRequest, ) -> Result<C2PendingRequest, C2ControlError>

pub async fn request( &self, route: NodeRoute, request: NodeRequest, ) -> Result<RoutedNodeResponse, C2ControlError>

pub async fn request_with_correlation( &self, route: NodeRoute, request: NodeRequest, correlation_id: Option<String>, ) -> Result<RoutedNodeResponse, C2ControlError>

pub async fn request_until( &self, route: NodeRoute, request: NodeRequest, budget: Duration, ) -> Result<RoutedNodeResponse, C2ControlError>

Self::request, but for a caller willing to wait out a physical reconnect: on Err(C2ControlError::ConnectionLost { .. }) or Err(C2ControlError::Closed) for a request NodeRequest::is_replay_safe admits, awaits the link-state watch (Self::link_state_receiver) turning C2LinkState::Connected again – bounded by whatever of budget is left – then resends, up to [REQUEST_UNTIL_MAX_ATTEMPTS] physical attempts total for this one logical call. A request is_replay_safe refuses, or any failure other than those two, returns on the very first attempt exactly like request does – this method changes nothing about request’s own behaviour or about control_owner’s pending-request failure handling; the retry lives here, one layer up.

budget == Duration::ZERO degenerates to exactly one attempt (the first elapsed-budget check trips before any wait), so a caller that wants request’s existing all-or-nothing semantics can pass that instead of calling request directly.

Trait Implementations

impl Clone for C2ReconnectingHandle

fn clone(&self) -> Self

Returns a duplicate of the value.

fn clone_from(&mut self, source: &Self)

Performs copy-assignment from source.