Handler for Proton messaging events. More...
#include <messaging_handler.hpp>
Public Member Functions | |
virtual void | on_container_start (container &) |
The container event loop is starting. | |
virtual void | on_container_stop (container &) |
The container event loop is stopping. | |
virtual void | on_message (delivery &, message &) |
A message is received. | |
virtual void | on_sendable (sender &) |
A message can be sent. | |
virtual void | on_transport_open (transport &) |
The underlying network transport is open. | |
virtual void | on_transport_close (transport &) |
The final event for a connection: there will be no more reconnect attempts and no more event functions. | |
virtual void | on_transport_error (transport &) |
Unexpected disconnect, transport::error() provides details; if reconnect_options are enabled there may be an automatic re-connect. | |
virtual void | on_connection_open (connection &) |
The remote peer opened the connection: called once on initial open, and again on each successful automatic re-connect (see reconnect_options). | |
virtual void | on_connection_close (connection &) |
The remote peer closed the connection. | |
virtual void | on_connection_error (connection &) |
The remote peer indicated a fatal error, connection::error() provides details. | |
virtual void | on_session_open (session &) |
The remote peer opened the session. | |
virtual void | on_session_close (session &) |
The remote peer closed the session. | |
virtual void | on_session_error (session &) |
The remote peer closed the session with an error condition. | |
virtual void | on_receiver_open (receiver &) |
The remote peer opened the link. | |
virtual void | on_receiver_detach (receiver &) |
The remote peer detached the link. | |
virtual void | on_receiver_close (receiver &) |
The remote peer closed the link. | |
virtual void | on_receiver_error (receiver &) |
The remote peer closed the link with an error condition. | |
virtual void | on_sender_open (sender &) |
The remote peer opened the link. | |
virtual void | on_sender_detach (sender &) |
The remote peer detached the link. | |
virtual void | on_sender_close (sender &) |
The remote peer closed the link. | |
virtual void | on_sender_error (sender &) |
The remote peer closed the link with an error condition. | |
virtual void | on_tracker_accept (tracker &) |
The receiving peer accepted a transfer. | |
virtual void | on_tracker_reject (tracker &) |
The receiving peer rejected a transfer. | |
virtual void | on_tracker_release (tracker &) |
The receiving peer released a transfer. | |
virtual void | on_tracker_settle (tracker &) |
The receiving peer settled a transfer. | |
virtual void | on_delivery_settle (delivery &) |
The sending peer settled a transfer. | |
virtual void | on_sender_drain_start (sender &) |
**Unsettled API** - The receiving peer has requested a drain of remaining credit. | |
virtual void | on_receiver_drain_finish (receiver &) |
**Unsettled API** - The credit outstanding at the time of the drain request has been consumed or returned. | |
virtual void | on_connection_wake (connection &) |
**Unsettled API** - An event that can be triggered from another thread. | |
virtual void | on_error (const error_condition &) |
Fallback error handling. |
Handler for Proton messaging events.
Subclass and override the event-handling member functions.
Thread-safety**: A thread-safe handler can use the objects passed as arguments, and other objects belonging to the same proton::connection. It *must not* use objects belonging to a different connection. See Multi-threading and proton::work_queue for safe ways to communicate between connections. Thread-safe handlers can also be used in single-threaded code.
Single-threaded only**: An application is single-threaded if it calls container::run() exactly once, and only makes make proton calls from handler functions. Single-threaded handler functions can use objects belonging to another connection, but *must* call connection::wake() on the other connection before returning. Such a handler is not thread-safe.
### Overview of connection life-cycle and automatic re-connect
See the documentation of individual event functions for more details.
on_connection_open() - The remote peer opened the connection: called once on initial open, and again on each successful automatic re-connect (see reconnect_options).
on_transport_error() -Unexpected disconnect, transport::error() provides details; if reconnect_options are enabled there may be an automatic re-connect.
on_connection_error() - The remote peer indicated a fatal error, connection::error() provides details.
on_connection_close() - The remote peer closed the connection.
on_transport_close() - The final event for a connection: there will be no more reconnect attempts and no more event functions.
virtual void on_container_start | ( | container & | ) | [virtual] |
The container event loop is starting.
This is the first event received after calling `containerrun()`.
virtual void on_container_stop | ( | container & | ) | [virtual] |
virtual void on_transport_close | ( | transport & | ) | [virtual] |
The final event for a connection: there will be no more reconnect attempts and no more event functions.
Called exactly once regardless of how many times the connection was re-connected, whether it failed due to transport or connection errors or was closed without error.
This is a good place to do per-connection clean-up. transport::connection() is always valid at this point.
virtual void on_transport_error | ( | transport & | ) | [virtual] |
Unexpected disconnect, transport::error() provides details; if reconnect_options are enabled there may be an automatic re-connect.
If there is a successful automatic reconnect, on_connection_open() will be called again.
transport::connection() is always valid at this point.
Calling connection::close() will cancel automatic re-connect and force the transport to close immediately, see on_transport_close()
virtual void on_connection_open | ( | connection & | ) | [virtual] |
The remote peer opened the connection: called once on initial open, and again on each successful automatic re-connect (see reconnect_options).
connection::reconnected() is false for the initial call, true for any subsequent calls due to reconnect.
if (!c.reconnected()) { Do initial per-connection setup here. Open initial senders/receivers if needed. } else { Handle a successful reconnect here. NOTE: Don't re-open active senders/receivers }
For a server on_connection_open() is called when a client attempts to open a connection. The server can call connection::open() to accept or connection::close(const error_condition&) to reject.
virtual void on_connection_close | ( | connection & | ) | [virtual] |
The remote peer closed the connection.
If there was a connection error, this is called immediately after on_connection_error() and connection::error() is set
virtual void on_connection_error | ( | connection & | ) | [virtual] |
The remote peer indicated a fatal error, connection::error() provides details.
A connection error is an application problem, for example an illegal action or a lack of resources. There will be no automatic re-connect attempts, on_connection_close() will be called immediately after.
virtual void on_sender_drain_start | ( | sender & | ) | [virtual] |
**Unsettled API** - The receiving peer has requested a drain of remaining credit.
virtual void on_receiver_drain_finish | ( | receiver & | ) | [virtual] |
**Unsettled API** - The credit outstanding at the time of the drain request has been consumed or returned.
virtual void on_connection_wake | ( | connection & | ) | [virtual] |
**Unsettled API** - An event that can be triggered from another thread.
This event is triggered by a call to `connectionwake()`. It is used to notify the application that something needs attention.
Thread-safety** - The application handler and the triggering thread must use some form of thread-safe state or communication to tell the handler what it needs to do. See `protonwork_queue` for an easier way to execute code safely in the handler thread.