[TUHS] seismo/uunet uucp

[TUHS] seismo/uunet uucp

[Lyndon Nerenberg (VE7TFX/VE6BBM)]

Does the source to the Rick Adams version of UUCP still exist anywhere?

--lyndon

[TUHS] Re: seismo/uunet uucp

[Clem Cole]

[-- Attachment #1: Type: text/plain, Size: 593 bytes --]

What is particular are you looking for? (I could be wrong here) but I
never kept it separately as but I had thought that Honeyman picked up
Rick's changes for later versions of honey-dan-ber  which I believe the
last revision was part of SVR4.  The last UUCP I ran was an SVR3-based
version, but by the time of SVR4, I had stopped any active UUCP work since
I had real IP network access.
ᐧ

On Tue, Jul 2, 2024 at 10:26 AM Lyndon Nerenberg (VE7TFX/VE6BBM) <
lyndon@orthanc.ca> wrote:

> Does the source to the Rick Adams version of UUCP still exist anywhere?
>
> --lyndon
>

[-- Attachment #2: Type: text/html, Size: 1245 bytes --]

[TUHS] Re: seismo/uunet uucp

[Lyndon Nerenberg (VE7TFX/VE6BBM)]

Clem Cole writes:

> What is particular are you looking for? (I could be wrong here) but I
> never kept it separately as but I had thought that Honeyman picked up
> Rick's changes for later versions of honey-dan-ber

What I'm particulary interested in is the last version of the
original (not HDB) uucp that they ran on SunOS (or whatever followed
-- I forget what OS they switched to afterwards).

My curiosity is to see what changes he made vs. the V7/BSD versions.
But I'm also looking for a well debugged version of the of the
pre-HDB version to run on my own machines.  With all the stupidity
in SMTP mail these days, a bunch of us have been setting up private
UUCP links to bypass all the middlebox nonsense and DMARC idiocy.
I was never a fan of the HDB configuration style, so I'm looking
for a rock solid implementation of Old School UUCP, and I figure
the seiso/uunet code is as bullet proof as it gets.

With that I can glue on TLS and we're good to go.

--lyndon

[TUHS] Re: seismo/uunet uucp

[Rich Salz]

[-- Attachment #1: Type: text/plain, Size: 169 bytes --]

Almost definitely, anything Seismo/UUNET/Rick did would be passed back to
CSRG.

You're going to glue TLS to UUCP?  You know it's interactive (two way)
protocol, right?

[-- Attachment #2: Type: text/html, Size: 239 bytes --]

[TUHS] Re: seismo/uunet uucp

[Lyndon Nerenberg (VE7TFX/VE6BBM)]

Rich Salz writes:

> Almost definitely, anything Seismo/UUNET/Rick did would be passed back to
> CSRG.

But did it?  I don't remember seeing it in any of the source trees
I was monkeying with.  Not that that means much.  But I also don't
remember ever seeing a non-gnu UUCP associated with any post-4.2 
release.  Thus leading back to my original question ...

> You're going to glue TLS to UUCP?  You know it's interactive (two way)
> protocol, right?

Yup.  It's just wrapping the connection with TLS, ala https vs http.
People have been doing it for decades (e.g. suucp in more tha a few
/etc/services files).  With HDB you can arrange this by doing some
contortions in the Dialers file that make it call out to 'openssl
s_client ...' IIRC.  It has been a looong time ...

--lyndon

[TUHS] Re: seismo/uunet uucp

[Grant Taylor via TUHS]

On 7/2/24 12:16 PM, Lyndon Nerenberg (VE7TFX/VE6BBM) wrote:
> Yup.  It's just wrapping the connection with TLS, ala https vs http. 
> People have been doing it for decades (e.g. suucp in more tha a few 
> /etc/services files).  With HDB you can arrange this by doing some 
> contortions in the Dialers file that make it call out to 'openssl 
> s_client ...' IIRC.  It has been a looong time ...

Does the UUCP that you're working with have any concept of -> support 
TCP based connections?  Or are you relying on external programs to dial 
TCP / TLS connections and similarly something else to host TCP / TLS 
connections and launch UUCP?



-- 
Grant. . . .
unix || die

[TUHS] Re: seismo/uunet uucp

[Rich Salz]

[-- Attachment #1: Type: text/plain, Size: 187 bytes --]

>
> > Almost definitely, anything Seismo/UUNET/Rick did would be passed back to
> > CSRG.
>
> But did it?
>

And how do you know S/U/R did any changes?

Shrug.  Not my hobby.  Good luck.

[-- Attachment #2: Type: text/html, Size: 449 bytes --]

[TUHS] Re: seismo/uunet uucp

[arnold]

"Lyndon Nerenberg (VE7TFX/VE6BBM)" <lyndon@orthanc.ca> wrote:

> Clem Cole writes:
>
> > What is particular are you looking for? (I could be wrong here) but I
> > never kept it separately as but I had thought that Honeyman picked up
> > Rick's changes for later versions of honey-dan-ber
>
> What I'm particulary interested in is the last version of the
> original (not HDB) uucp that they ran on SunOS (or whatever followed
> -- I forget what OS they switched to afterwards).
>
> My curiosity is to see what changes he made vs. the V7/BSD versions.
> But I'm also looking for a well debugged version of the of the
> pre-HDB version to run on my own machines.  With all the stupidity
> in SMTP mail these days, a bunch of us have been setting up private
> UUCP links to bypass all the middlebox nonsense and DMARC idiocy.
> I was never a fan of the HDB configuration style, so I'm looking
> for a rock solid implementation of Old School UUCP, and I figure
> the seiso/uunet code is as bullet proof as it gets.
>
> With that I can glue on TLS and we're good to go.
>
> --lyndon

Maybe check out GNU UUCP? I think it could use old style
configuration files.

Arnold

[TUHS] Re: seismo/uunet uucp

[Lyndon Nerenberg (VE7TFX/VE6BBM)]

arnold@skeeve.com writes:

> Maybe check out GNU UUCP? I think it could use old style
> configuration files.

Come on, this is TUHS :-)  I'm trying to resurrect the original
code, not some newcomer clone :-)

--lyndon

[TUHS] Re: seismo/uunet uucp

[arnold]

"Lyndon Nerenberg (VE7TFX/VE6BBM)" <lyndon@orthanc.ca> wrote:

> arnold@skeeve.com writes:
>
> > Maybe check out GNU UUCP? I think it could use old style
> > configuration files.
>
> Come on, this is TUHS :-)  I'm trying to resurrect the original
> code, not some newcomer clone :-)
>
> --lyndon

Oh, sorry. I merely thought that you wanted to get something done
and working.

[TUHS] Re: seismo/uunet uucp

[Jeremy C. Reed]

On Tue, 2 Jul 2024, Lyndon Nerenberg (VE7TFX/VE6BBM) wrote:

> Does the source to the Rick Adams version of UUCP still exist anywhere?

See the CSRG SCCS archives.
4.3BSD has 5.6     85/06/24
4.4BSD has 8.1     93/06/06

2.9BSD-30aug85.gz  has changes from rick@seismo.ARPA
June 19, 1985

[TUHS] Re: seismo/uunet uucp

[Clem Cole]

[-- Attachment #1: Type: text/plain, Size: 6888 bytes --]

Lyndon,

I just untar'ed "file6" of the 4.3BSD *.tap file from TUHS, which has the
usr.bin sources (which include UUCP).

This is clearly a BSD version but has been influenced by Honey-Dan-Ber.

It seems to have the 't' protocol. [I did not look to see if it has the 'e'
protocol, which was the earlier TCP over ethernet that I released from
Masscomp about four years earlier and is also in UUNET Netnews Archives.]
Given the comments about supporting Tymnet's NonPeak services, I feel
pretty confident that this is post Rich Adam's work.

Clem
--- CHANGES FILE ---
List of Changes to UUCP
CHANGES 5.7 86/02/12

Added support for Eunice.

Added new dialers:
Racal Vadic 212
Racal Vadic 811 dialer with 831 adaptor
Racal Vadic 820 dialer with 831 adaptor
Racal Vadic MACS, 811 dialer with 831 adaptor
Racal Vadic MACS, 820 dialer with 831 adaptor
Dec DF112
4.2BSD style networking on top of tcp/ip ('t' protocol)
X.25/PAD support ('f' protocol)
Novation
Penril
Hayes 2400 Smartmodem
Concord Data Systems CDS 224
ATT 2224 2400 baud modem

Running uucico with debugging on requires read access to L.sys.

If "NOSTRANGERS" is defined in uucp.h, the remote site
must be in you L.sys or the call will be rejected.

Lock files may be kept in a subdirectory if desired.

STST files are kept in a subdirectory.

CORRUPT subdirectory contains corrupted C. and X. files that could
not be processed. (Instead of just exiting)

You can specify a maximum grade to send either on the command line (-gX)
or in the L.sys file (Any/C|Evening will only send class C [usually
mail] or higher during the day and will send everything in the evening)
See UUAIDS/L.sys for examples.

L.sys (and any of the files in lib/uucp) can contain comments by
putting a # as the first character on a line. Lines may be
continued by placing a \ as the last character of the line.

-R flag reverses role. (Lets the remote system be master first instead
of slave)

-L flag only calls "local" sites. Local sites are those sites having
one of LOCAL,TCP or DIRECT in the ACU field of L.sys.

If /etc/nologin is present (usually created by a graceful shutdown),
uucico and uuxqt will gracefully exit instead of getting killed
off when the system goes down.

Does an exponential backoff on retry time if call fails instead
of always waiting the default 5 minutes. The default may be
overridden by adding ",TIME" to the time field in L.sys. e.g.
"seismo Any,2" will use a default retry time of 2 minutes.

If uucico receives a SIGFPE while running, it will toggle debugging
on and off.

Better status messages provided for uustat.

New program uuq to give more decriptive information on status of
jobs in uucp queue.

Don't send files to remote system if it is returning out of
temp space error.

Correctly does the closing hangup sequence.

condevs.c was broken into a file for each dialer in the
directory aculib for much easier maintenance.

Only try at most TRYCALLS to dial a site instead of one try
for each dialer (lost big on systems with many dialers)

Add ABORT sequence to the expect/send sequence so don't have
to wait for timeout if can't get through dataswitch. e.g.
noao Evening ACU 1200 6021234565 "" \d\r CLASS NOAOUUCP ABORT Down GO \d\r
ogin:-\b-ogin: uucplogin word: uucppassword
will only call noao in the evening (evening is defined by the phone rates).
It will expect nothing and then wait 1 second (\d) and send a carriage
return.
Look for CLASS, then send NOAOUUCP. From then on, if it sees the word Down
before finishing logging in, it will hang up immediately. In the mean time,
it looks for GO and if it sees it, delays 1 second and sends a CR. Looks
for ogin:, etc. This abort sequence is very useful if you must go through a
dataswitch to get to the computer.

The time field in L.sys now handles "Evening" and "Night" in addition
to Any, Mo,Tu,We,Th,Fr,Sa,Su. Evening and Night are defined to
be when the phone rates are cheaper.
Evening = Wk1700-0800|Sa|Su
Night = Any2300-0800|Sa|Su0800-1700

The expect/send code now supports:
\s space
\d delay 1 second
\r carriage return with no linefeed
\b break
\c don't send a CR after these characters
\xxx the octal character xxx (e.g. \s == \040

L-devices now handles "chat" scripts (like HoneyDanber) to get
through local port selectors and smart modems more easily
without mucking up every line of L.sys. See UUAIDs/L-devices for
details.

The 'g'  protocol code was cleaned up a lot and is now almost
readable.

If you need a parity other than even (the default) to login to
another system, you can change it in L.sys by putting in a
sequence like "" P_ZERO (expect nothing, send zero parity).
Odd Parity P_ODD
Even Parity P_EVEN
Zero Parity P_ZERO
One Parity P_ONE

If DONTCOPY is defined in uucp.h, uucp will not make a copy
of the source file by default. (This is the way System 3 does it).

If an X. request fails, the notification is returned to the
originator of the request instead of "uucp" on the previous
system.

The man pages are actually accurate!

If LOGBYSITE is defined, uucp logging is done with
a log file per site instead of one LOGFILE. (Like Honey DanBer does)

There is a new file (optional) L.aliases that makes life simpler when
a site changes it's name. uucp, uux, uucico, etc all check it so
when a site is renamed (e.g convex <- parsec) all you have to do is add
an entry in L.aliases of the form:
newname oldname

uucico will not try and resend files it has already sent (when the
files are specified in one C. file)

Incorporated Bill Sebok's code to dial in and out on the same modem.
NOTE: acucntrl is heavily Vax/4.xbsd specific and will require work to
run on any other system.

For compatibility with Honey DanBer, in the Date fields of L.sys,
| was changed to , (| is supported, but not encouraged)
, was changed to ; (to allow , to be the date seperator)

For Honey DanBer compatibility, the Grade flag is now passed as
-vgrade=X instead of the old -pX

Don't truncate site names to 7 characters (truncate to 14 if
anyone gets that absurd) for HDB compatibility.  L.aliases may
be used to map host with longer names in L.sys to 7 character
names that some hosts send.  Entries should be

fullname 7-char-name

You can specify a time for the expect send sequences with ~ instead
of getting the default MAXMSGTIME. E.g.
system Any ACU 1200 1234567 ogin~20-\r-ogin~10-\b-ogin user password pw
will look for ogin for 20 seconds, send CR, look for ogin for 10
seconds, send a BREAK and look for ogin for MAXMSGTIME seconds

Added code to support GTEs PC Pursuit service. It's mainly the handling of
the
dialback they use.

Added time "NonPeak" for Tymnet/Telenet services that charge lower
rates from 6pm-7am M-F and Sat & Sun.
ᐧ
ᐧ

[-- Attachment #2: Type: text/html, Size: 9120 bytes --]

[TUHS] Re: seismo/uunet uucp

[Lyndon Nerenberg (VE7TFX/VE6BBM)]

Clem Cole writes:

> I just untar'ed "file6" of the 4.3BSD *.tap file from TUHS, which has the
> usr.bin sources (which include UUCP).

Clem, that appears to be what I'm looking for. I thought 4.3 had
already switched to gnu, so I never thought to look there.

Thanks!

--lyndon

[TUHS] Re: seismo/uunet uucp

[arnold]

"Lyndon Nerenberg (VE7TFX/VE6BBM)" <lyndon@orthanc.ca> wrote:

> Clem Cole writes:
>
> > I just untar'ed "file6" of the 4.3BSD *.tap file from TUHS, which has the
> > usr.bin sources (which include UUCP).
>
> Clem, that appears to be what I'm looking for. I thought 4.3 had
> already switched to gnu, so I never thought to look there.
>
> Thanks!
>
> --lyndon

As a matter of history, I don't think GNU UUCP existed that early.
I could be wrong, but I don't think I am.

Arnold

[TUHS] Re: seismo/uunet uucp

[Alexander Schreiber]

On Tue, Jul 02, 2024 at 12:32:23PM -0400, Rich Salz wrote:
> Almost definitely, anything Seismo/UUNET/Rick did would be passed back to
> CSRG.
> 
> You're going to glue TLS to UUCP?  You know it's interactive (two way)
> protocol, right?

Wrapping UUCP in stunnel isn't exactly rocket science - and my mail
has been flowing through this for more than 20 years ;-)

Kind regards,
           Alex.
-- 
"Opportunity is missed by most people because it is dressed in overalls and
 looks like work."                                      -- Thomas A. Edison