Interesting. If message accumulation on sending side waits till EOL is pressed - or some other indication of end of message is entered - then network wouldn't

Interesting. If message accumulation on sending side waits till EOL is pressed - or some other indication of end of message is entered - then network wouldn't actually see "lo" - they wouldn't be sent before the crash. If input would be sent character-by-character, then first "l" would be sent, then "o", then system would crash. In this case the first message would be "l". This wouldn't be TCP connection - that would require some handshakes beforehand, so more messages back and forth even before "l" is sent. There were protocols before TCP; simplest way is to assume it's direct IP. So, first point-to-point, routed internetworked message - first IP packed reached destination across networks - seemed to contain payload "l". What's wrong here?