UUCP over X.25

[Carl S. Gutekunst]

>Is there something like that?

Sure. Half the world is still using UUCP or X.25 as its primary transport
medium for UNIX E-mail and news.

If you are running end-to-end between host X.25 implementations (no external
PAD), and the implementation has some way for UUCP to easy place calls, then
it is possible to run plain old 'g' protocol over X.25. It's can be expensive,
though; 'g' data packets are 70 bytes long (two segments!), and all those
uucico RR packets tend to come back one packet per. The uucico RR delays can
also kill throughput.

To cut costs, and to run properly across PADs that are not 8-bit transparent,
Piet Beertema and Robert Elz wrote a thing called 'f' protocol. The source is
public domain; it has been widely integrated in non-AT&T Unixes. (Peter Honey-
man actually gave the AT&T UNIX folks a version of HoneyDanBer that had 'f'
protocol in it. AT&T ripped it out, claiming they already had their own UUCP
or X.25 protocol, called 'x'.) The 'f' protocol assumes that the link is
"mostly error free," and not 8-bit transparent; it quotes all 8-bit and non-
printing characters (good for PADs, bad for costs), and does not send any
data acknowledgements until the very end of the file (a big win).

The afore mentioned 'x' protocol is garbage. Don't waste your time.

Of course, there are also lots of ways to run higher-level protocols over X.25
(including SNDCF and ISO transport), and if your X.25 can use one of those you
can accomplish something similar. But it adds a lot of overhead, and requires
both ends to be running the same high-level protocol; I don't think that is
what you were asking.... 

Note that many international sites have now abandoned X.25 in favor of dial-up
lines and Telebit TrailBlazers. They are considerably faster than 9600-baud
international X.25, and about a factor of 10 cheaper. Not as realiable (calls
don't always go through to some countries), but at these prices, who cares? 

<csg>