Abstract interface for network transport backends. More...
#include <NetworkBackend.hpp>
Inheritance diagram for MayaFlux::Core::INetworkBackend: Collaboration diagram for MayaFlux::Core::INetworkBackend:Public Member Functions | |
| virtual void | close_endpoint (uint64_t endpoint_id)=0 |
| Close an endpoint and release its resources. | |
| virtual EndpointState | get_endpoint_state (uint64_t endpoint_id) const =0 |
| Query the current state of an endpoint. | |
| virtual std::vector< EndpointInfo > | get_endpoints () const =0 |
| List all endpoints currently managed by this backend. | |
| virtual std::string | get_name () const =0 |
| virtual NetworkTransport | get_transport () const =0 |
| virtual std::string | get_version () const =0 |
| virtual bool | initialize ()=0 |
| Initialise backend resources (sockets, SHM segments, etc.) | |
| virtual bool | is_initialized () const =0 |
| virtual bool | is_running () const =0 |
| virtual uint64_t | open_endpoint (const EndpointInfo &info)=0 |
| Open a new endpoint. | |
| virtual bool | send (uint64_t endpoint_id, const uint8_t *data, size_t size)=0 |
| Send data through an endpoint. | |
| virtual bool | send_to (uint64_t endpoint_id, const uint8_t *data, size_t size, const std::string &address, uint16_t port)=0 |
| Send data to a specific address through an endpoint. | |
| virtual void | set_receive_callback (NetworkReceiveCallback callback)=0 |
| Register the receive callback. | |
| virtual void | set_state_callback (EndpointStateCallback callback)=0 |
| Register the endpoint state change callback. | |
| virtual void | shutdown ()=0 |
| Release all resources, close all endpoints. | |
| virtual void | start ()=0 |
| Start receive threads and accept connections. | |
| virtual void | stop ()=0 |
| Stop receive threads without releasing resources. | |
| virtual | ~INetworkBackend ()=default |
Detailed Description
Abstract interface for network transport backends.
Follows the same lifecycle and structural pattern as IInputBackend: initialize() -> start() -> [running] -> stop() -> shutdown()
Each concrete implementation (UDP, TCP, SharedMemory) owns sockets or shared segments, manages background receive threads, and exposes a uniform endpoint model. The NetworkSubsystem owns backends and routes high-level operations through this interface.
Bidirectional: sends and receives, manages persistent endpoints rather than devices. This is the core structural difference from IInputBackend.
Backends are deliberately synchronous and callback-based. They know nothing about coroutines. The coroutine surface lives in NetworkService and is implemented by NetworkSubsystem using callbacks internally to resume suspended coroutines. This keeps backends simple (pure OS socket wrappers) and avoids coupling transport code to the Vruta/Kriya stack.
Endpoint lifecycle: open_endpoint(info) -> endpoint_id send(endpoint_id, data, size) [receive via callback] close_endpoint(endpoint_id)
Future transport backends (WebRTC, GStreamer, etc.) implement this same interface. Protocol-level orchestration above raw transport (codec pipelines, session management, jitter buffering) belongs in a future Portal::Network layer, not here.
Definition at line 64 of file NetworkBackend.hpp.
The documentation for this class was generated from the following file:
