Oh, I ran into that when developing inetb. When threads terminated abnormally (like, I forcibly killed them) it made a mess in net_server, which inevitably needed to be restarted (but at least it could be restarted). I had to figure out a way to gracefully tell threads to go away even when they...