It is not trivial to run online
nncp-caller commands over the serial
link, because it is link without built-in error detection. For efficient
usage you have to use some kind of window-sliding error correction
protocol, like Kermit, ZMODEM, UUCP’s g-protocol and similar well known
However TCP is already satisfying and existing protocol for the same purposes. So it would be more easier to bring up the IP interconnection and use TCP over it. Most distributions already have PPP protocol out-of-box.
Ordinary man page for
ppp in FreeBSD is enough for being able
to setup it:
/usr/local/bin/ppplogin: #!/bin/sh exec /usr/sbin/ppp -direct incoming
/etc/ttys: ttyU0 "/usr/libexec/getty 3wire.115200" vt100 on
incomingPPP profile, allowing authenticated
pppuser to log in:
/etc/ppp/ppp.conf: incoming: allow ppp disable deflate pred1 mppe protocomp vjcomp ipcp dns enable lqr # null-modem cables might not have carrier detection /etc/ppp/ppp.secret: ppp PASSWORD
/etc/ppp/ppp.conf: outgoing: set device /dev/cuaU0 set speed 115200 set dial set login set authname ppp set authkey PASSWORD disable deflate pred1 mppe protocomp vjcomp ipcp dns enable lqr
That configuration does not negotiate any kind of IPv4 addresses, routing or DNS servers. Also all compression is turned off, because NNCP’s traffic is encrypted and uncompressible. Only IPV6CP will negotiate IPv6 link-local addresses, on which you can run multicast discovered daemons for simplicity.