> It sure did last time I read the RFC; either end can send a FIN packet
> and continue to receive data in the other direction "forever", until
> the other end sends a FIN or until the local TCP client insists on
> closing down the read half of the connection.

That would imply there's some more complex problem with the BSD telnet 
implementation that the System V implementation gets right, since it should
be able to send a FIN on EOF then wait for the server to close.