🚀 go-pugleaf

I need some help.

I am moving files between computers. Big files, many megabytes. The
transfer initiated from within a UNIX application, running on the host
where the files are created, and the files are sent to a remote
computer. When the receiving computer also runs UNIX and talks TCP/IP,
I know what to do, and it works. This goes with local networks, leased
lines and dialed-up telephone lines through protocols like ethernet,
SLIP, and PPP. It would even work with avian carriers (RFC 1149).

I am a UNIX addict and a young one. I remember only few computers not
having full networking (as in 4.2BSD UNIX or DEC TOPS-20).

Now, to my surprise, some computers don't have native TCP/IP. I am
talking Apple Macintosh here, and maybe IBM MS-DOS PC. Suddenly, I
need to send files to these small computers at remote sites. To match
their low price, I must exclude every carrier other than dialed-up
telephone lines.

Note that I, having the large machine, need to send files to these
smaller computers. There are plenty of kermit-like programs that allow
home computer owners to connect to a larger system, but this is of no
use to me. Here, the small computer needs to answer the call.

I could run some non-native TCP/IP. On the Macintosh I could use
Apple's MacTCP and MacSLIP. However, from information on USENET, I
understand that neither MacSLIP nor its competitors (InterCon and
Versaterm) will listen for incoming modem calls.  MacTCP and MacSLIP
simply makes for yet another kermit-like system. I don't know, but I
guess the same goes for MS-DOS computers.

I have been playing around with thoughts of implementing my own
protocols, but I know far too little about Macintosh or PC network
programming for this. It would take more time than I have got, first
to learn to program these single-process computers and then to design
my own protocol implementations.

There is one use of small computers where they are communication
servers and that is Bulletin Board Systems. I have seen a few, but
they are simply too awkward for my needs. They assume a user at the
modem. I have a UNIX application at my modem and only maybe a user at
the application.

I have been experimenting with UUCP, a protocol from the PDP 11 age.
My UNIX, as all UNIXes, still features support for this protocol and,
to my happy surprise, there are public domain implementations for both
the IBM PC and Apple Macintosh. I have tried UUPC 3.0 for the Mac and
it works. On my Hewlett Packard UNIX (HP-UX 7.0) I use the built-in
HDB UUCP.

I am not all happy with the UUCP solution. It is very complicated to
set up. On the UNIX side, it puts my job in its own queue and makes me
believe everything is alright. When something fails, I receive a mail.
It is hard for my UNIX application to know the status of the transfer.
Is it still in progress? Is it done? Did it fail?

Again, the UNIX UUCP assumes a user commanding it. There doesn't seem
to be a programmer's interface to UUCP. I need my UNIX application to
manage the transfer. The end user only knows this application, she is
neither a UUCP expert nor a UNIX wizard.

Tell me, did you ever have this problem? Did you solve it? How? Are
there any books I should read?

-- 
Lars Aronsson, SECTRA AB, Teknikringen 2, 583 30 Linkoping, Sweden