I'm considering xmlrpc++ as a candidate implementation to use for a client-server application. I'd like to know two things before I start off:
- Does the server spawn a new worker thread for each requested rpc ?
- If not, what happens when a new rpc arrives while the server is busy processing a previous rpc ? Is the new one stored in some queue or is it discarded ?
Thanks,
George
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
- No, the server currently does not spawn a new thread per RPC. In a future release that may be an option. In some other threads in this forum there are discussions and even some code showing how to subclass XmlRpcServer and XmlRpcServerConnection to run requests in separate threads.
- The method XmlRpcServer::bindAndListen takes 2 arguments, the port to bind and listen on, and the size of the backlog queue. This queue is for stacking connect requests while the server is busy.
If the backlog queue is full when a new connect request is received, that client will not be able to connect and will receive an error (usually connection refused).
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
Hello,
I'm considering xmlrpc++ as a candidate implementation to use for a client-server application. I'd like to know two things before I start off:
- Does the server spawn a new worker thread for each requested rpc ?
- If not, what happens when a new rpc arrives while the server is busy processing a previous rpc ? Is the new one stored in some queue or is it discarded ?
Thanks,
George
- No, the server currently does not spawn a new thread per RPC. In a future release that may be an option. In some other threads in this forum there are discussions and even some code showing how to subclass XmlRpcServer and XmlRpcServerConnection to run requests in separate threads.
- The method XmlRpcServer::bindAndListen takes 2 arguments, the port to bind and listen on, and the size of the backlog queue. This queue is for stacking connect requests while the server is busy.
If the backlog queue is full when a new connect request is received, that client will not be able to connect and will receive an error (usually connection refused).