Folks,
I've been doing some experimenting with RoverLog networking, and I can reproduce issues folks have been reported with very sluggish performance. It's nasty.
These happen when messages are sent to nonexistent IP peer nodes with the current UDP scheme.
The old scheme used TCP, but did not create a permanent socket between nodes. It opened and closed a socket with each message. This was TERRIBLE because of the amount of IP traffic it created.
The UDP scheme is TERRIBLE if a node disappears temporarily or never existed. There is a long hang with each network attempt which I did not anticipate (it's UDP after all).
So I am working on a scheme that uses TCP but creates permanent sockets between nodes and grooms and maintains these sockets. I think this will be a good scheme long term. It'll take me a little while to get this going, but this should really fix issues where networks are spotty or where the settings represent a full multi-op system but some stations are not powered up.
Please bear with me as I revamp RoverLog networking for the next release.
Thanks,
Tom, N1MU.