Please report new issues athttps://github.com/DOCGroup
In TAO_ECG_UDP_Receiver::handle_input, if an error is returned, -1,from call to recv() on the dgram socket then the handle_input routine returns -1. This causes the reactor to remove the handler for that socket. Additionally, no error messages are printed or otherwise indicated to the process. This causes the application to suddenly and silently go def on the multicast event channel, at least for events originating from other, federated event channels. This occurs in 2 points in the routine, 1 for reading the header and the other at reading the entire datagram. In our case, removal of the handle effectively ends participation in the simulation (RTI). This behavior should be user configurable. The arriving packet may contain a checksum error, which on some OS's returns -1 and sets errno to EAGAIN (EWOULDBLOCK). Other OS's handle return 0 on this error which leads to platform specific behavior for the EC UDP gateway. In the case of returning -1 from this routine, use of perror would be preferable to at least see that some error had occurred, perhaps again being configurable.
Re-assign to tao-support, I personally won't have any time to work on it.
Accepting for tao-support
FYI, in the latest version of the code now an error is printed.