Proposal: Usepi, a retrohobbyist UUCP network for the PiDP-*community

[Jon Brase]

Before Google Groups, before the Web,  In days of yore, when minicomputers roamed the earth, and ARPANET took its first shaky steps onto dry land, when neither snow nor rain nor heat nor gloom of night could stay swift sysadmins from lugging sacks of e-mail on horseback down 1200 baud lines ------

*record scratch

Who am I kidding? I remember the 80s (barely), they weren't *that* long ago!

In any case, back when DEC was still selling the PDP-11, there was a network called Usenet. I'm sure many here remember it (actually, it still exists, but since the Web came into being, there have been enough commercial services offering Usenet access to just about anyone that it's mostly spam and other unpleasantries these days). For those who don't, machines connected to USENET didn't have the always-on high-bandwidth connections we enjoy today. Traffic on Usenet was generally (in the early days, at least) transferred over scheduled dial-up calls that a machine made to, or received from, its neighbors, often made at night, when phone rates were lower. Usenet was built on the UUCP protocol, and traffic on the network was primarily e-mail and newsgroups (a precursor to things like web forums and Google Groups).

Is anybody interested in starting a hobbyist UUCP network for the PiDP-* community? I imagine the scope of discussion for a network being primarily retrocomputing, with a focus on the PiDP-* line.

[Clem Cole]

Jon,

     Please feel free to try.  Note that I'm CC'ing a few of the folks involved, whom I encourage to chime in.

     I am seriously not trying to rain on your parade but I am warning you it has been tried before. It seems like a cool idea.  But setting it up was a lot of work but failed to stick.  By using The Unix Heritage Society (TUHS) as his base, Warren did actually create a retro Usenet a few years before the 50th UNIX anniversary (managed as part of the 50th-anniversary mailing list). To his credit, he managed to get many of the people involved with the original USENET, to set up simh based systems reclaiming their old node names.   It lasted for a few weeks/months, but after it was set up, it sort of died.  I actually think the structure may be still there, but nothing really runs over it.

IMO the problem was that really was no need for it anymore.  Mary Ann's wonderful efforts of creating the original USENET User groups have continued to exist but run now over the Internet directly, plus were pulled into other news subsystems that originally forked from it.   Of course, email flows better over the same.  UUCP's rise was really because a low-cost IP connection was not available at the time, and a connection to the Internet cost about $250K a year (if you get one from DoD or later CSNET).   So running UUCP over WE212 or Racal-3 triple modem and a few years later Telebit Trailblazer or eventually X.25 links to UUNET.com 'just worked' and anyone could get them.  The cost was your telephone bill and find a local phone call friend to toss your data too.  The long haul stuff was being supported by 'big sites' like AT&T (ihnp4), DEC (decvax), Tektronix (teklabs), HP (hplabs), with UCB (ucbvax) being the major ARPAnet/Internet gateway.  FWIW: The whole reason we at the USENIX Association funded Rick Adam's UUNET proposal was that the membership and board saw the original scheme could not last with DEC spending $600K/yr on phone charges (although at one point, someone at AT&T realized they generated an average of 10-20 downstream long-distance calls for each one ihnp4 placed - so it was a financial gain for them).

So, the use case/need for a UUCP based network sort of fell away when the personal original ASDL and later cheap broadband became available and small ISPs birthed to supply connectivity.  (Sadly, cable - Comcast in particular - killed most of them off - but that's a different story).   The only use case left for uucico is as a transfer scheme over simulated serial lines between a host and simh based PDP-11 for old OS's that lack LAN (Ethernet/TCP) support.  Frankly, is someone wrote virtual shared drive between the host/simh (like the E11 simulator and most VM managers) support it is not even clear that would useful.  [To be honest, I just use virtual tape when I need to move things back and forth].

Anyway - that's my thoughts.  If you can get folks excited, go for it.   But I fear you will find what we did, you are filling a need that long ago left us and it's not clear the nostalgia/historical connect has much value for that piece.

Best wishes,

Clem 

--
You received this message because you are subscribed to the Google Groups "[PiDP-11]" group.
To unsubscribe from this group and stop receiving emails from it, send an email to pidp-11+u...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/pidp-11/15c4a365-30c2-4d99-93e6-1edae3110d61o%40googlegroups.com.

[mi...@mikeharpe.com]

I set up and managed a couple of UUCP connections back in the day. I had no love for it. Needlessly complex setup and management. I was unimpressed. Admittedly this was on System V on the AT&T 3B2 so there's that. The only thing worse was getting printers working.

What I want is a paper tape reader/punch. Those were fun.

Plus, I'll bet you'd be amazed at how many people on this list don't have a modem anymore. Mine got purged years ago. If you're not going to do it with a modem, you're not really doing UUCP.

Mike Harpe

[Steven A. Falco]

On 7/4/20 10:40 AM, mi...@mikeharpe.com wrote:
> I set up and managed a couple of UUCP connections back in the day. I had no love for it. Needlessly complex setup and management. I was unimpressed. Admittedly this was on System V on the AT&T 3B2 so there's that. The only thing worse was getting printers working.
>
> What I want is a paper tape reader/punch. Those were fun.
>
> Plus, I'll bet you'd be amazed at how many people on this list don't have a modem anymore. Mine got purged years ago. If you're not going to do it with a modem, you're not really doing UUCP.

There's also the problem that a lot of land lines are VoIP these days - not at all modem-friendly, at least at the higher baud rates.

Steve

[Adam Thornton]

If we rebuild uunet, we may as well do BITNET while we're at it.

Don't listen to me, though.  I'm currently having trouble getting v7 to communicate over UUCP to its host on the Pi.  Which is irritating me, because once I have a decent file-transfer system to and from v7, since the s editor is working, it's very close to being a usable system for me.

--
You received this message because you are subscribed to the Google Groups "[PiDP-11]" group.
To unsubscribe from this group and stop receiving emails from it, send an email to pidp-11+u...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/pidp-11/ce3db330-8c19-6117-3ce9-186a2faa6326%40gmail.com.

[Mike Katz]

I can echo that sentiment.  Getting UUCP up and running was a bitch.  Even the Fidonet to UUCP gateway was a hassle.

--

You received this message because you are subscribed to the Google Groups "[PiDP-11]" group.
To unsubscribe from this group and stop receiving emails from it, send an email to pidp-11+u...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/pidp-11/d00fcca1-cd98-4bf7-a486-6d1d889dada0n%40googlegroups.com.

[Mike Katz]

I would recommend doing something like UUCP over internet rather than land lines.

We could even used DDNS to overcome most of us not having static IP addresses.

--
You received this message because you are subscribed to the Google Groups "[PiDP-11]" group.
To unsubscribe from this group and stop receiving emails from it, send an email to pidp-11+u...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/pidp-11/15c4a365-30c2-4d99-93e6-1edae3110d61o%40googlegroups.com.

[Adam Thornton]

But if you're going to do UUCP-over-internet....just use NNTP.  It works fine.  Then you get a flood-fill distribution system with hierarchical channels basically for free, and a rich set of tools to work with.

No one says you have to use the usual Usenet groups.  There's nothing stopping you from creating a private hierarchy and using it.  I don't know if there were ever TLS extensions to NNTP but you could always use stunnel or similar to wrap your transport, if you cared.

UUCP's remaining role, in so far as I am concerned, is to be able to transfer files to and from systems lacking a TCP/IP stack.  For me, right now, the one I care about is v7 Unix.  Oddly, using BITNET in that role is sort of supplanted by BITNET's transport mechanism.  That is, with Hercules, I can just create punch files and send them to the virtual card reader, and receive them inside VM/370 (at some point, resurrecting the v7 port to the IBM 370 by way of VM/370 is on my wouldn't-it-be-lovely list).

To view this discussion on the web visit https://groups.google.com/d/msgid/pidp-11/569060ad-94f3-62c2-8e3f-a9faf3642a72%40gmail.com.

[Jon Brase]

On Saturday, July 4, 2020 at 1:28:17 PM UTC-5, Mike Katz wrote:

I would recommend doing something like UUCP over internet rather than land lines.

Oh, certainly. I wasn't thinking of doing anything else. Probably we'd have a script of some sort connect to a simh telnet port (since simh can't establish outbound connections), and then the script would function as a modem for "dialing" an ssh connection to a remote site (to avoid sending plaintext telnet over the open internet), and then the username used for the ssh connection would need to have its shell set to a script that would telnet to the appropriate simh port.

[Jon Brase]

On Saturday, July 4, 2020 at 12:55:56 PM UTC-5, Adam Thornton wrote:

If we rebuild uunet, we may as well do BITNET while we're at it.

Don't listen to me, though.  I'm currently having trouble getting v7 to communicate over UUCP to its host on the Pi.  Which is irritating me, because once I have a decent file-transfer system to and from v7, since the s editor is working, it's very close to being a usable system for me.

I have a setup guide for UUCP on v7 at https://github.com/jwbrase/pdp11-tools/tree/master/howtos , which may be of use. Or, if you've been following that, I'd certainly love to have feedback on any trouble you've run into.

The main problem I ran into is that the system that comes with the PiDP-11 OS kit is configured to use DC-11 serial, but the DC-11 isn't 8-bit clean, and v7 UUCP needs 8-bit clean lines (the protocols later developed for UUCP on narrower bit-widths didn't exist yet). So a kernel recompile is necessary to switch to using the DZ-11 for serial instead, and the process for the recompile is about a third of my guide.

[Clem Cole]

On Sat, Jul 4, 2020 at 3:41 PM Adam Thornton <atho...@gmail.com> wrote:

UUCP's remaining role, in so far as I am concerned, is to be able to transfer files to and from systems lacking a TCP/IP stack.  

Exactly - I'll take a shot at poking at it on my system tomorrow if I can (family stuff with the next couple of generations later today).

Another thought/FYI: Bruce Borden and I have been trying to see if we can find the old UNET sources (I was their first customer for the same back-in-day - it was 3Com's first product).  The V7 based UNET code is an old TCP stack by today's standards, and originally only talked to the original Xerox 3MBIT board, and then the first 3Com ethernet board (and later Steve Glaser and I wrote the original HyperChannel driver for it to talk to the CDC Boxes and the VMS).  But Telnet and FTP worked fine to V7.  At Tektronix, we only ran on it on 11/70 class systems, we never tried to squeeze it not 11/40 class machines, but Bruce, Greg Shaw, Metcalfe, and the rest their team at 3Com must have at some point.  The problem is we can not find complete sources so far, but if we do find them, Warren will get them for the archives ... sigh.

[Adam Thornton]

I followed it, mostly.  I already had defined the DZ-11 and set it up for 16 lines, and since I've been using it for terminal access, I know the driver is pretty much working.

I set it to 8 bits, and that worked fine.

The problem is that when I send from the Pi side, I get:

uucico v7 - (2020-07-03 21:56:55.93 28773) Calling system v7 (port TCP)
uucico v7 - (2020-07-03 21:57:03.45 28773) Login successful
uucico v7 - (2020-07-03 21:57:05.13 28773) Handshake successful (protocol 'g' sending packet/window 64/3 receiving 64/7)
uucico v7 adam (2020-07-03 21:57:12.40 28773) Sending /home/adam/git/simh/sim_scsi.h (6780 bytes)
uucico v7 adam (2020-07-03 22:01:40.15 28773) ERROR: Timed out waiting for packet
uucico v7 - (2020-07-03 22:01:40.15 28773) Protocol 'g' packets: sent 86, resent 6, received 1
uucico v7 - (2020-07-03 22:01:40.15 28773) Errors: header 2, checksum 0, order 0, remote rejects 0
uucico v7 - (2020-07-03 22:01:41.03 28773) Call complete (281 seconds 5440 bytes 19 bps)

That is, the initial logon and protocol handshake succeeds, but then I never receive some acknowledgement I am waiting for on the v7 end, and nothing ever appears in /usr/spool/uucp/public .

v7 has considerably less debugging available, but when I initiate a transfer from its end, I likewise never see anything arrive on the pi end.

Adam

--

You received this message because you are subscribed to the Google Groups "[PiDP-11]" group.
To unsubscribe from this group and stop receiving emails from it, send an email to pidp-11+u...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/pidp-11/911d1fed-d4de-4268-8788-7314ce55ba68o%40googlegroups.com.

[Adam Thornton]

Also, telnetting to the terminal port as the pi-uucp user behaves about like I would expect; at least, there's something there and it is clearly running uucico.

adam@pibplus:~/git/simh $ telnet -8 localhost 1107
Trying ::1...
Connected to localhost.
Escape character is '^]'.


Connected to the PDP-11 simulator DZ device, line 0

lo�i�:�
login: pi-uucp
Password:
Shere

(I don't know why the first login prompt has high bits set; maybe a speed issue?  If I hit Enter it looks fine from there.)

I defined MYNAME to "v7" in uucp.h:

# cat uucp.h
#include "stdio.h"
        /*  some system names  */
#define MYNAME          "v7"

L.sys has:

Format of these lines is:        sysname time to call : device needed : speed
16bitpi None ACU 300 555-5555 "" ""

The name of the pi on the far end is indeed 16bitpi:

system v7
myname 16bitpi
time any 1
max-retries 10
call-login pi-uucp
call-password s33kr1t
port v7-dz
address localhost
chat "" \d\d\r ogin: \d\L word: \P

alternate v7-alt
port v7-dc-pipe

[Jon Brase]

What does your simh boot.ini file look like for the v7 system? I recall that I ran into some character mangling issues due to incorrect attach flags for the DZ device.

Jon Brase

[Adam Thornton]

echo
echo Type boot and at the : prompt type in hp(0,0)unix
echo
set cpu 11/70
#set xu enable
#set xu mac=06-DE-AD-BE-EF-01
#att xu tap:tap0
set cpu 2M
set cpu idle
set rp0 rp06
att rp0 rp06-0.disk
set rp1 rp06
att rp1 rp06-1.disk
set dz en
set dz lines=16
set tto 7b
att -m dz 1107
#set dz 7B
# UUCP needs 8 bits
set dz 8b
d cpu 2000 042102

boot rp0

[Jon Brase]

Ah. I've found it. I neglected to include the following in the setup guide:

set dz lines = 16
att dz -m 1107
att dz -am line=n,1108;notelnet
set dz 8b

Where "n" is whatever line you want to dedicate to UUCP.

The "notelnet" bit is the really critical part. Taylor UUCP connects to the specified port raw, and simh by default tries to talk telnet back to it. This works for long enough to log in, and then something gets corrupted and the connection hangs.

I think what happened is I thought I had it running and finalized the guide, and then I realized that I was getting far too many dropped calls and mangled files, but then neglected to update the guide when I found the fix.

I keep the UUCP login on a separate line so that the other lines can still speak telnet protocol for human logins.

[Adam Thornton]

It's painfully slow, but this time it is sending and receiving the queued files.  "sum" agrees that the contents are the same.  Thank you!

I think at this point I have a decent screen editor and a way to get files on and off the system that is less janky than "paste slowly" through iTerm2.

Adam

[Jon Brase]

On Saturday, July 4, 2020 at 9:23:11 AM UTC-5, Clem Cole wrote:

Jon,

     Please feel free to try.  Note that I'm CC'ing a few of the folks involved, whom I encourage to chime in.

     I am seriously not trying to rain on your parade but I am warning you it has been tried before. It seems like a cool idea.  But setting it up was a lot of work but failed to stick.  By using The Unix Heritage Society (TUHS) as his base, Warren did actually create a retro Usenet a few years before the 50th UNIX anniversary (managed as part of the 50th-anniversary mailing list). To his credit, he managed to get many of the people involved with the original USENET, to set up simh based systems reclaiming their old node names.   It lasted for a few weeks/months, but after it was set up, it sort of died.  I actually think the structure may be still there, but nothing really runs over it.

My idea is to have a UUCP network, but not to try and recreate the exact connection graph of the historical Usenet (or recreate the connection graph at all, exactly or not). I feel doing so would (and likely did) involve a lot of extra setup work to find volunteers to run each system, and then each volunteer might very well be running several systems that they're not really doing anything with other than administering, and the extra work would likely dampen enthusiasm for all involved. I'm thinking something like each person that wants to participate establishes links with the ~5 people closest to them in meatspace, and only brings in the systems that they want to (rather than just spinning up simh instances to fill a slot), using their own hostnames, rather than "OK, somebody has to be ucbvax, might as well be me". That way people are more personally invested in the chunk of the network they're running, which I think is important for a nostalgia project.

[Jon Brase]

On Saturday, July 4, 2020 at 2:41:33 PM UTC-5, Adam Thornton wrote:

But if you're going to do UUCP-over-internet....just use NNTP.  

One could just as easily argue "If you're going to run 2.11 BSD/v7 on modern hardware, just use Ubuntu". :-)

No one says you have to use the usual Usenet groups.  There's nothing stopping you from creating a private hierarchy and using it.

 

Which is basically where "Usepi" in the thread title came from: me trying to think up a good name for a top-level hierarchy for the network. "UsePiDP" sounded a bit clunky, "PiNet" could easily be used for Rasperry Pi projects that don't go anywhere near retrocomputing, let alone the PDP-11 or Usenet, and I think I mulled over a few other names that I can't remember. I think I nixed something on length, it might have been the aforementioned "UsePiDP", or it might have been something longer.  "Usepi" is two syllables, like Usenet, the "Use" part draws a connection to Usenet, which connects to retrocomputing and especially to retrocomputing with the types of systems that were on Usenet in its heyday, and the "Pi" part draws a connection to the Raspberry Pi.

My proposal would be to start out with one top-level hierarchy, "usepi", one second-level hierarchy "pidp", and five groups:

usepi.test         For new systems to verify that they're set up properly

usepi.meta       Discussion around usepi itself, including proposals for new groups

usepi.pidp.8     Discussion around the PiDP-8. I'm a bit doubtful that a netnews suite would fit on a PDP-8, but that doesn't prevent it from being discussed over netnews.

usepi.pidp.11   Discussion around the PiDP-11.

usepi.pidp.10   Discussion around the PiDP-10.

If we wanted to be cute, we could have "dp" instead of "pidp" for our second level hierarchy:

usepi.dp.11

Likely early expansions would be:

usepi.retro.*          Hierarchy for retrocomputing in general. Or, this might be our second level hierarchy, with "pidp" being a third-level hierarchy under it.

usepi.pidp.*.$OS  Groups for the OSes that the various pidp's can run.

Of course, there would be nothing preventing participants from being part of other networks and carrying their newsgroups: if there are at least some systems from the 50th anniversary project still running, I could imagine usepi being carried on some of them, and I could imagine some systems on usepi getting feeds from that network for its groups. About the only network I'd rule out is the wretched hive of spam and spamandspam that is Usenet itself these days. Even then, the wonderfully decentralized nature of UUCP networking means that  a participant *could* carry Usenet if they could stand the signal to noise ratio and didn't get snowed under by the volume (they'd probably need modern server hardware, and lots of it). As long as they didn't try to feed Usenet proper to their neighbors, the neighbors would have no reason to know or care.

[Adam Thornton]

Fair enough.  If usepi gets off the ground, I will be happy to be a node.

If you carry a _very_ carefully curated Usenet feed, it's possible to use Outer Usenet in the modern era, and the actual traffic is low enough that UUCP is quite plausible.  The following seems to work OK for me, although I am only actually reading about half of them (and some of them may well have gone years without posts):

rec.arts.int-fiction
rec.games.int-fiction
rec.games.roguelike.nethack
alt.fan.warlord
alt.folklore.computers
alt.lang.intercal
alt.music.tom-waits
alt.peeves
alt.sysadmin.recovery

Adam

--

You received this message because you are subscribed to the Google Groups "[PiDP-11]" group.
To unsubscribe from this group and stop receiving emails from it, send an email to pidp-11+u...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/pidp-11/2b61e3a5-ce8c-4c9c-9492-ee9b54af1b98o%40googlegroups.com.

[will...@gmail.com]

Wait, wha?1 You got s working on v7? Would you mind sharing the code? I spent a half day on it a few weeks back, but got bogged down. I would appreciate it.

Will

[jon....@gmail.com]

Sorry, I missed this back when you posted it.

My HowTos for getting UUCP running are on my github: https://github.com/jwbrase/pdp11-tools/tree/master/howtos

[Warren Toomey]

Clem mentioned that, a few years back I tried to recreate a UUCP/Usenet (CNews) environment. This was with simulated Vaxen running 4.3BSD and using TCP connections as simulated land lines. All the code is at https://github.com/DoctorWkt/4bsd-uucp. We got to about 30 machines in the network. At the time, I also got a 2.11BSD SimH system working with UUCP and CNews, and it was connected to the UUCP network. So, this is definitely doable but, as Clem mentioned, it all fizzled out due to lack of interest.

And I did see someone who had also set up UUCP and CNews on a 2.11BSD image, somewhere on Github but, alas, I can't find it now. That would make it much easier :-)

Cheers, Warren

[jon....@gmail.com]

On Tuesday, June 14, 2022 at 6:07:40 PM UTC-5 warren...@gmail.com wrote:

Clem mentioned that, a few years back I tried to recreate a UUCP/Usenet (CNews) environment. This was with simulated Vaxen running 4.3BSD and using TCP connections as simulated land lines. All the code is at https://github.com/DoctorWkt/4bsd-uucp. We got to about 30 machines in the network. At the time, I also got a 2.11BSD SimH system working with UUCP and CNews, and it was connected to the UUCP network. So, this is definitely doable but, as Clem mentioned, it all fizzled out due to lack of interest.

I understand that that effort involved trying to recreate the actual historical layout of Usenet at a more or less particular point in time? This is something a bit different in philosophy from that: As stated upthread; I don't want people that join to feel obliged to duplicate particular historic systems, but rather just to have a network that people can join their own existing retrocomputing systems to (probably mostly PiDP-11s at first, since that's the community I'm pitching it in).

 

And I did see someone who had also set up UUCP and CNews on a 2.11BSD image, somewhere on Github but, alas, I can't find it now. That would make it much easier :-)

I can't find anybody else that's got a repository for a 2.11 BSD build of CNews on github, but at https://github.com/jwbrase/CNews-PiDP-11 , you'll find my modifications to the CNews sources to get them to build on 2.11 BSD, and at https://github.com/jwbrase/pdp11-tools/tree/master/howtos , you'll find howtos for getting UUCP, sendmail, and CNews running on 2.11 BSD (as well as for barebones UUCP on v7 Unix). I deliberately have *not* provided an image: I want people to be able to start with the stock PiDP-11 2.11 BSD image, follow my howtos using my modifications to the CNews sources, and build a system with working CNews; that way people that have already added other stuff to their 2.11 BSD image don't have to choose between their existing image and the CNews image, they can just add CNews to the existing image.

[Adam Thornton]

I'm in a private Usenet hierarchy (which works pretty well), and I would suggest that creating a similar one for retro machines attached to this new network, accessible by invitation/request only, would be a much better idea than trying to hook it up to the cesspool that remains of public Usenet.

Adam

[Bob Flanders]

IIRC, the MOSHIX channel (about mainframes) has a BITNET working now...

https://www.youtube.com/watch?v=SpUgWt9OWB0&t=1070s

Maybe we could join this?

To view this discussion on the web visit https://groups.google.com/d/msgid/pidp-11/4609dbde-fcd6-428e-927a-6098339df4d8n%40googlegroups.com.

[jon....@gmail.com]

My idea has always been for a private hierarchy (even if modern public Usenet weren't a cesspit, I'm sure it would overwhelm a PDP-11, even an emulated one running on hardware considerably beefier than a Raspberry Pi).

Even what Warren was running was a private hierarchy: it just attempted to recreate the historical connection graph of Usenet at some point in the mid-80s (or that's the impression I get, Warren can correct me).

I certainly would be interested in hearing about whatever hierarchy you're involved with.

[jon....@gmail.com]

Same concept, different implementation, and a different set of typical machines that it typically ran on. This seems to imply that there was at least some interconnectivity between USENET and BITNET, but I have no idea if the set of specific machines with historical connectivity to both networks included any PDP-11s, or whether any PDP-11 OS had software stacks for both protocols (such that it could have been theoretically possible for a PDP-11 to be an interconnect point between the two networks, whether that possibility was ever realized or not), or indeed, whether a BITNET stack existed for any PDP-11 OS at all.

Certainly getting BITNET up and running on a PiDP-11 would be an interesting thing to do, assuming that a suitable BITNET stack exists, as would getting UUCP and BITNET up on the same machine (assuming a suitable pair of software stacks exists), but the point of this thread (at least until someone can post a howto on getting BITNET running, or better yet, on getting a UUCP-BITNET interconnect node running) is to discuss putting together a UUCP network.

[Adam Thornton]

On Jun 19, 2022, at 12:28 AM, jon....@gmail.com <jon....@gmail.com> wrote:

Same concept, different implementation, and a different set of typical machines that it typically ran on. This seems to imply that there was at least some interconnectivity between USENET and BITNET, but I have no idea if the set of specific machines with historical connectivity to both networks included any PDP-11s, or whether any PDP-11 OS had software stacks for both protocols (such that it could have been theoretically possible for a PDP-11 to be an interconnect point between the two networks, whether that possibility was ever realized or not), or indeed, whether a BITNET stack existed for any PDP-11 OS at all.

Certainly getting BITNET up and running on a PiDP-11 would be an interesting thing to do, assuming that a suitable BITNET stack exists, as would getting UUCP and BITNET up on the same machine (assuming a suitable pair of software stacks exists), but the point of this thread (at least until someone can post a howto on getting BITNET running, or better yet, on getting a UUCP-BITNET interconnect node running) is to discuss putting together a UUCP network.

BITNET <-> Usenet gateways certainly existed in the late 80s/early 90s at least.

I think BITNET used NJE as its underlying transport.  There was a BSD (-?) -licensed NJE-for-Unix-ish machines that we used at Sine Nomine in the mid-2000s, and I think I remember porting it to use then-modern autotools.

So in theory this ought to be a simple matter of programming.  Use NJE to send/receive things from/to a BITNET spool, and then pick those things out, perform whatever translation you need to (in many cases, presumably, EBCDIC->ASCII) on them, transform them into Usenet posts, and dump 'em on the news spool.

I think you can do it record-at-a-time so memory constraints are likely not problematic.  

Adam

[Heinz-Bernd Eggenstein]

So back to Usepi/USENET...

I've now manged to get my P(i)DP-11 (under 2.11 BSD) to exchange files via uucp (over TCP) with a VAX-11/780 (in SIMH

 of course) under 4.3BSD within my home network (directly, without installing uucp on any of the Raspberry Pis) . That was a bit of work, mostly because of

a) missing directories (and wrong user/group ownership) under /usr/spool/uucp on the VAX / 4.3BSD side

b) me not knowing the first thing about uucp and having to consult some tutorials first, some were already mentioned in this thread.

Anyway, now that I begin to understand the whole thing a bit better, I wonder how we would set up a USENET via TCP/IP these days without inadvertently compromising the security of our home networks.

If I get this right, at least one node in the network would have to accept "calls" from other nodes, and I guess this means there would have to be a port opened in your firewall (e.g. 540) for this.

Needless to say, any open port that is exposed to the internet gets bombarded by port scanners and bots these days. As for security, I guess the main concern is not so much that by exploiting some ancient hack someone could take control of your PDP-11 inside SIMH, but that somehow it would be possible to break out of SIMH and do stuff in our home networks.

Has anyone  given thought about how to make this secure, without too much hassle for the users (no complicated VPN solution etc I guess). I guess if that dial-in server would exist in a dedicated network with no other hosts to be compromised even in the event of a breach, this could be done without too many headaches?

Cheers

HB  

[Adam Thornton]

>
> Needless to say, any open port that is exposed to the internet gets bombarded by port scanners and bots these days. As for security, I guess the main concern is not so much that by exploiting some ancient hack someone could take control of your PDP-11 inside SIMH, but that somehow it would be possible to break out of SIMH and do stuff in our home networks.
>
> Has anyone given thought about how to make this secure, without too much hassle for the users (no complicated VPN solution etc I guess). I guess if that dial-in server would exist in a dedicated network with no other hosts to be compromised even in the event of a breach, this could be done without too many headaches?

I mean, my approach has been just not to worry about it. I make a number of ancient systems available over the web. If you can a) figure out how to break out of the emulated machine into simh, b) get from simh to a shell, c) get from that shell on one of my emulator machines (mostly Raspberry Pis) to some machine on my network with data I really care about on it, d) break into that machine, e) defeat the passphrases on the encryption on the stuff I care about to take it...then you care enough that it was probably easier to just break into my house and beat me with a rubber hose until I gave you what you wanted.

Like, I put my credit cards in other people's POS machines multiple times a day on average; I have commercial relationships with dozens of companies that clearly store my information unencrypted since it keeps leaking and I will have credit monitoring for life. There are easier ways to steal my stuff than by going through my emulators.

Adam

[Heinz-Bernd Eggenstein]

Fair enough. It's one thing to accept the risk as a personal, informed decision, but one has to be careful when you publish a recipe for how to join all this, to not understate the risks to less infomed minds.

Anyway,  I guess we have a volunteer for a "dial-in" server then :-) ??

What we would need then is to have a few people execute the HowTos for setting up UUCP and CNews on their PiDP-11s under BSD2.11 to start with, set up the message exchange schedules and then from there on, create usepi.comp.infosystems.usepi  or  some such  and then discuss the rest there :-).

Count me in! Who else want's to join?

HB

[Adam Thornton]

> On Jul 1, 2022, at 3:53 AM, 'Heinz-Bernd Eggenstein' via [PiDP-11] <pid...@googlegroups.com> wrote:
>
> Fair enough. It's one thing to accept the risk as a personal, informed decision, but one has to be careful when you publish a recipe for how to join all this, to not understate the risks to less infomed minds.
> Anyway, I guess we have a volunteer for a "dial-in" server then :-) ??

Yeah, when USEPi or whatever gets off the ground, I'm in, and I'll probably build an HNET gateway for it because honestly how could I not. Maybe I'll use some of the spare capacity to run a telnet-accessible 90s BBS.

>
> What we would need then is to have a few people execute the HowTos for setting up UUCP and CNews on their PiDP-11s under BSD2.11 to start with, set up the message exchange schedules and then from there on, create usepi.comp.infosystems.usepi or some such and then discuss the rest there :-).
>
> Count me in! Who else want's to join?

Adam

[jon....@gmail.com]

My thought on security has generally been that we'd do something like a "modem" or "phone exchange" script on the Pi that telnets into whatever port the dialout line is listening on and listens for a "dial" command, which would be part of the UUCP call script (or, for systems running 2.11 BSD, where there's a TCP stack, I suppose the process could just listen for a telnet connection from the 2.11 BSD instance). When the dial command was received, the modem process would then make a key-based ssh connection to the appropriate remote PiDP-11, where ssh would be configured to run a particular command on login for the username used for the connection, namely to telnet to an incoming port configured to accept UUCP connections (the attach settings needed for UUCP to receive a clean line are a bit different than those you want for an interactive login).

[Clem Cole]

As the author of the original E protocol for running UUCP over ethernet, I'll offer a few other tricks courtesy of those of us scared from years gone by.

Using ssh and telnet are fine -- another suggestion.   Create different login's >>names/password<< other than uucp that have slightly different behaviors.   For instance, it could fork traditional uucico without having to use Chesson's g protocol.  

1.) in the 'chat script' negotiate a different protocol since your connection is over a reliable TCP connection, g is just overhead.

2.) Have a link to uucico itself -- called uucico-e (or uucico-t which is the other one that came later). Then have the hosts that is telnet/ssh in login as uucp-e and as part of the login, uucico-e 'knows' to only offer e (or t) during the protocol negotiation.

Suggestion -- try to find the original ORA 'Managing UUCP' book.   IIRC Dave D wrote it, but I've forgotten.  Many (most) of the kinds of things you might want to do are addressed in that book, so it will save you hours of relearning.   There may be a PDF Of in the wild, I never looked, but it should be available on the used market for a small $s [shipping likely more expensive than anything].

Recently the TUHS mailing has been discussing things young hackers have no idea / would never remember.  I'll add one here - what is called in UUCP the uucico(8) 'chat' script - birthed a famous [and quite useful in hindsight] UNIX utility - expect(1).  If you look at the uucico(8) source you will see why expect(1) was called that.  Such a simple idea, but until it was originally added to uucico(8) it was never formalized.   The folks at NIST broke it out as a separate program with its own language/command system.

ᐧ

[Jon Brase]

Jul 1, 2022 15:14:12 Clem Cole <cl...@ccc.com>:

> Suggestion -- try to find the original ORA 'Managing UUCP' book.   IIRC Dave D wrote it, but I've forgotten.  Many (most) of the kinds of things you might want to do are addressed in that book, so it will save you hours of relearning.   There may be a PDF Of in the wild, I never looked, but it should be available on the used market for a small $s [shipping likely more expensive than anything].

So there appear to be two different editions, a 1990 edition that deals with HDB UUCP, traditional v2 UUCP, Xenix UUCP, and a fair bit of Usenet stuff not related to UUCP proper, and a 1996 edition with a focus on UUCP proper that doesn't cover v2 UUCP but does cover Taylor UUCP.

The former would have been indispensable for my effort bringing up UUCP on v7 Unix; I had to track down some early v2 UUCP documentation by (IIRC) Kernighan that assumed a lot of existing UUCP and v7 knowledge on the part of the reader that (young whippersnapper that I am) I had to figure out on my own.

The latter will be useful to those that want to include their host Pi, or a Linux system on their network, among the UUCP neighbors for their guest system, but TBH, Taylor UUCP's info pages are fairly good, so if one were to decide to buy only one of the two, I'd originally go for the 1990 edition.

I'll definitely have to check them out.

> Recently the TUHS mailing has been discussing things young hackers have no idea / would never remember.  I'll add one here - what is called in UUCP the uucico(8) 'chat' script - birthed a famous [and quite useful in hindsight] UNIX utility - expect(1).

I'm not sure if I knew there was a direct connection, but I had definitely noticed the similarity.

[Clem Cole]

1st. I don’t think bwk wrote anything about UUCP.  Iirc Dan N.  wrote much of the original doc as part of the V7 release.  Generally it was “trust the source Luke” since we all had it in those days when we ran it on our PDP 11s.

2nd I’m pretty sure the 1990 vs of Managing UUCP is the second version.  The first version was V7 and BSD only in the early mid 80s. It was one of Tim’s first books.  HDB had not been written yet.    I’ve got a copy somewhere as I was one of the original reviewers    As I said my memory is that Dale wrote it but I’m not sure of that. FWIW  It might have had a different name.  

--

Sent from a handheld expect more typos than usual

[Adam Thornton]

I'm surprised there even is a 1996 UUCP book. By 1990 the writing was already on the wall that TCP/IP was the present-and-future. By 1996...everybody had TCP/IP, even if only via SLIP or PPP, and you just copied stuff around online.

Adam

[Jon Brase]

Jul 1, 2022 19:53:38 Clem Cole <cl...@ccc.com>:

> 1st. I don’t think bwk wrote anything about UUCP.  Iirc Dan N.  wrote much of the original doc as part of the V7 release.  Generally it was “trust the source Luke” since we all had it in those days when we ran it on our PDP 11s.

It was "UUCP Implementation Description", indeed, by D.A. Nowitz, not bwk as I had thought. It appears to have been part of the v7 manual but not indexed as a man page, so I probably confused it with some document of similar origins that was written by bwk.

> 2nd I’m pretty sure the 1990 vs of Managing UUCP is the second version.  The first version was V7 and BSD only in the early mid 80s. It was one of Tim’s first books.  HDB had not been written yet.    I’ve got a copy somewhere as I was one of the original reviewers    As I said my memory is that Dale wrote it but I’m not sure of that. FWIW  It might have had a different name.  

Whether 1990 was the second or the first, it was the 1990 and 1996 versions that came up most immediately in a quick Google search.

The title for the 1996 version is a bit different, it's listed as written by Dale Dougherty.

The 1990 version is by Tim O'Reilly, and, actually, looking at it now, Amazon lists it as a Tenth Edition. The 1996 is a first edition with its different title, both are published by O'Reilly media.

1990 is "Managing UUCP and Usenet", 1996 is "Using and managing UUCP". It looks like 1996 is a successor with a narrower focus.

[Clem Cole]

Jon tx.  I thought Tim might have written the first edition (we shared a card table as our desk at Masscomp back in the day).  I do have memories of explaining to him how it worked and helping him get it set up at ORA in Cambridge when we loaned him a MC500 to get going (he originally had one connection- Masscomp).  But I was unsure if he wrote the first edition so I did not offer that.  However I was pretty sure Dale wrote the later version.