[TUHS] Berkeley CSRG Building

[TUHS] Berkeley CSRG Building

[Pete Wright via TUHS]

Hey there folks, I'm going to find myself in Berkeley for the day this 
week and was curious if the buildings where the CSRG group worked are 
still standing?  Thought it would be a fun place to visit after hacking 
on BSD for all these years.

Cheers,
-pete

-- 
Pete Wright
pete@nomadlogic.org

[TUHS] Re: Berkeley CSRG Building

[Erik E. Fair]

The key building is Evans Hall, a large, brutalist croncrete thing that is slated to be demolished & replaced for earthquake concerns.

	https://en.wikipedia.org/wiki/Evans_Hall_(UC_Berkeley)

CSRG's offices & machine room (including ARPANET IMP #78 when that was delivered; before that, ARPANET was connected to the Ingres PDP-11/45 (or 11/70?) in Cory Hall) was on the 4th floor, east side - windowed offices had a view to the Hearst Mining Circle and the Molecular Biology buildin - windowed offices had a view to the Hearst Mining Circle and the Molecular Biology building.

The CS department (College of Letters & Science) offices were on the 5th floor.

The main campus computer center was in the basement, but they charged by the CPU/second, so only classes used those systems - in my day (1980-1983), the computer center had a CDC 6400 and six or seven DEC PDP-11/70s. BerkNet (RS-232c long haul lines) connected the PDP-11s, and later the VAXen, though Ethernet supplanted it.

	https://en.wikipedia.org/wiki/Berknet

The other building of note is Cory Hall - that's where the EECS (College of Engineering) was headquartered, and the Cory Hall DEC PDP-11/70 was in a couple of rooms on the south side of the west entrance, with an additional terminal room just opposite on the north side.

	https://eecs.berkeley.edu/about/visiting/

I believe all of CS has since moved into Soda Hall on the other side of Hearst Avenue, which is built atop the site of UCB's former (decommissioned) nuclear reactor.

	https://en.wikipedia.org/wiki/TRIGA

The neighborhood has changed, but it looks like La Val's Pizza is still there on Euclid Avenue - one of the favored student hangouts in those days. The Top Dog sausage restaurant has closed its Hearst Avenue branch, but the location on Durant Avenue (southside of campus) is still there for additional local flavor. There were (and probably still are) many fine restaurants along Shattuck Avenue in downtown Berkeley and points north.

I also recommend the commanding view westward of the San Franciso Bay from the Lawrence Hall of Science (LHS) -

	https://lawrencehallofscience.org/visitors/directions-parking/

which was also used as the exterior of an AI computer control center for a dystopian movie:

	https://en.wikipedia.org/wiki/Colossus:_The_Forbin_Project

Some of that view could also be had in the 10th floor (west side) UCB Mathematics grad student lounge in Evans Hall, but I think that's no longer accessible.

	Erik Fair

[TUHS] Re: Berkeley CSRG Building

[Pete Wright via TUHS]

Amazing - thanks so much!
-pete
Sent from my iPhone

> On Aug 12, 2024, at 1:16 AM, Erik E. Fair <fair-tuhs@netbsd.org> wrote:
> 
> The key building is Evans Hall, a large, brutalist croncrete thing that is slated to be demolished & replaced for earthquake concerns.
> 
>    https://en.wikipedia.org/wiki/Evans_Hall_(UC_Berkeley)
> 
> CSRG's offices & machine room (including ARPANET IMP #78 when that was delivered; before that, ARPANET was connected to the Ingres PDP-11/45 (or 11/70?) in Cory Hall) was on the 4th floor, east side - windowed offices had a view to the Hearst Mining Circle and the Molecular Biology buildin - windowed offices had a view to the Hearst Mining Circle and the Molecular Biology building.
> 
> The CS department (College of Letters & Science) offices were on the 5th floor.
> 
> The main campus computer center was in the basement, but they charged by the CPU/second, so only classes used those systems - in my day (1980-1983), the computer center had a CDC 6400 and six or seven DEC PDP-11/70s. BerkNet (RS-232c long haul lines) connected the PDP-11s, and later the VAXen, though Ethernet supplanted it.
> 
>    https://en.wikipedia.org/wiki/Berknet
> 
> The other building of note is Cory Hall - that's where the EECS (College of Engineering) was headquartered, and the Cory Hall DEC PDP-11/70 was in a couple of rooms on the south side of the west entrance, with an additional terminal room just opposite on the north side.
> 
>    https://eecs.berkeley.edu/about/visiting/
> 
> I believe all of CS has since moved into Soda Hall on the other side of Hearst Avenue, which is built atop the site of UCB's former (decommissioned) nuclear reactor.
> 
>    https://en.wikipedia.org/wiki/TRIGA
> 
> The neighborhood has changed, but it looks like La Val's Pizza is still there on Euclid Avenue - one of the favored student hangouts in those days. The Top Dog sausage restaurant has closed its Hearst Avenue branch, but the location on Durant Avenue (southside of campus) is still there for additional local flavor. There were (and probably still are) many fine restaurants along Shattuck Avenue in downtown Berkeley and points north.
> 
> I also recommend the commanding view westward of the San Franciso Bay from the Lawrence Hall of Science (LHS) -
>    https://lawrencehallofscience.org/visitors/directions-parking/
> 
> which was also used as the exterior of an AI computer control center for a dystopian movie:
> 
>    https://en.wikipedia.org/wiki/Colossus:_The_Forbin_Project
> 
> Some of that view could also be had in the 10th floor (west side) UCB Mathematics grad student lounge in Evans Hall, but I think that's no longer accessible.
> 
>    Erik Fair

[TUHS] Re: Berkeley CSRG Building

[Al Kossow]

On 8/12/24 1:16 AM, Erik E. Fair wrote:

> The main campus computer center was in the basement, but they charged by the CPU/second
I made a similar pilgrimage when I moved out here in the mid 80s
I think the 6600 console was still in the hall, and there was a want ad for
a kernel hacker up on the wall from DEC WRL when they were bringing up Titan
I made a cold call there, but Smokey wasn't around (I was young and dumb!)

[TUHS] Re: Berkeley CSRG Building

[Clem Cole]

[-- Attachment #1: Type: text/plain, Size: 4427 bytes --]

Erik - great job - nothing to quibble about, here adding a little color.

On Mon, Aug 12, 2024 at 4:16 AM Erik E. Fair <fair-tuhs@netbsd.org> wrote:

> The key building is Evans Hall, a large, brutalist croncrete thing that is
> slated to be demolished & replaced for earthquake concerns.
>
>         https://en.wikipedia.org/wiki/Evans_Hall_(UC_Berkeley)
>
> ARPANET was connected to the Ingres PDP-11/45 (or 11/70?) in Cory Hall
>
Ing70 via a VDH interface to LBL was rather late in the ARPANET history.

UCB was supposed to have its own connection earlier, but for reasons I
never knew, it did not happen.   I understand that it was targeted to be
installed in the computer center, and UCB politics somehow waylayed it.
When the Ingres contract was let, the Ingres team got the VDH connection as
a part of it.   No other systems could connect to the ARPANET.  As Erik
mentioned, in '78 Eric Schimdt's MS thesis work was developing the RS-232C
9600 baud "Berk-Net" for three machines -- 2 in Evan and then a wire in the
steam tunnel that Bob Kridle pulled allowed them to connect the Cory
Hall70, which was the "student" system that was were the primary work
for 1BSD and 2BSD  occurred (ex/csh et al.).   Bob later ran a serial line
up the stairwells of Cory and picked up the Ingres and the first CAD
machine.  Kurt Shoens, who was the primary UCB Mail author, hacked the
Berknet support into his work.  One of his primary additions was removing
the "delivery" part of the mail into a separate program - that he called
"delivermail."

Eric Alman was the system administrator of Ing70, so Eric hacked
delivermail to pass email to and from the Berknet to the ARPANET.  All was
good until Ernie covax showed up and was connected to the UUCP net ;-)

Eric refers to that time as "the email format of the week" and sendmail was
created to allow him to more easily handle the different formats.  By then
there was the DARPA 733/822 format (user@host), Berknet (host:user), UUCP
(host1!host2...!hostn!user) being sources at UCB, as well as crap showing
up from the IBM Educational System, CSNET and various other places trying
be exchanged.


> The CS department (College of Letters & Science) offices were on the 5th
> floor.
>
Ah .. but CS was a division of EECS -- very interesting history about
that.  The EE Dept was considered top three in the country and #1 by many
ratings.  They already had a huge history of working with industry and had
created the Industrial Laison Office (ILO - which was how the UNIX BSD
tapes were distributed until the creation of CSRG and a factoid. The ILO
folks advised Fabry on how to run CSRG)).  When CS was created, it was
forked from the Math Dept. But, the Regents felt that if the CS was to have
a chance to become a top 10 program, it needed to be put under the auspices
of the EE Dept like MIT was -- and was attached to EE -- hence EECS.

>
> The main campus computer center was in the basement, but they charged by
> the CPU/second, so only classes used those systems - in my day (1980-1983),
> the computer center had a CDC 6400 and six or seven DEC PDP-11/70s.

I could be wrong - but ISTR, there was an IBM 360/370 system down there
also.

The 70s were used to teach CS courses such as the one I taught -- I think
Bart Miller, Mike Carey and I taught CS-40 a qtr or two after Erik was an
undergrad.



> BerkNet (RS-232c long haul lines) connected the PDP-11s

Actually, not at all, which was all the more amazing.  There were just CAT3
twisted pair with no conditioning.   When we had an electrical storm, it
was not usually to have the replace the 488/489 transcievers in the DHs or
DZs.  Kridle got pretty adept and putting machines sockets into the boards
and he kept a box of chips in his office.



> , and later the VAXen, though Ethernet supplanted it.
>
It was initially with 3Mbit Xerox boards. We were a little itchy about
that, so Bob got an optical coupler/repeater (I assume from Xerox, but I
don't know) in Cory's CAD machine room.   The 10M stuff came about two
years later, originally 3COM equipment with.  There was a LAN in the Ingres
machine room, another in the CAD, and another in CSRG's in Evans.   The UCB
CAD machine was on all three networks as it was in the middle.  That is why
Sam wrote the original routed stuff -- based on things he had seen at PARC
for PUP.

[-- Attachment #2: Type: text/html, Size: 7216 bytes --]

[TUHS] Re: Berkeley CSRG Building

[Luther Johnson]

[-- Attachment #1: Type: text/plain, Size: 5464 bytes --]

I really enjoy all the history and perspective. A phrase just popped in
to my head, that I think for me, sums up much of Unix evolution. And it
is something that not all Linux developers get, as many continue to
mutate Linux further and further away from traditional Unix ideas:

Necessity is the mother of *convention*.

I think what I wish more of the Linux community would understand, is
that most of Unix design style is what it is, in order to deal with the
above-stated reality, in a simple and practical way. Many times, when
faced with a choice that might seem arbitrary at the time, but because I
know that I never know, which choices will live on and have future
impacts, I think WWUD (what would Unix do) ?

On 08/12/2024 10:34 AM, Clem Cole wrote:
> Erik - great job - nothing to quibble about, here adding a little color.
>
> On Mon, Aug 12, 2024 at 4:16 AM Erik E. Fair <fair-tuhs@netbsd.org
> <mailto:fair-tuhs@netbsd.org>> wrote:
>
>     The key building is Evans Hall, a large, brutalist croncrete thing
>     that is slated to be demolished & replaced for earthquake concerns.
>
>     https://en.wikipedia.org/wiki/Evans_Hall_(UC_Berkeley)
>     <https://en.wikipedia.org/wiki/Evans_Hall_%28UC_Berkeley%29>
>
>     ARPANET was connected to the Ingres PDP-11/45 (or 11/70?) in Cory Hall
>
> Ing70 via a VDH interface to LBL was rather late in the ARPANET history.
>
> UCB was supposed to have its own connection earlier, but for reasons I
> never knew, it did not happen.   I understand that it was targeted to
> be installed in the computer center, and UCB politics somehow waylayed
> it.  When the Ingres contract was let, the Ingres team got the VDH
> connection as a part of it.  No other systems could connect to the
> ARPANET.  As Erik mentioned, in '78 Eric Schimdt's MS thesis work was
> developing the RS-232C 9600 baud "Berk-Net" for three machines -- 2 in
> Evan and then a wire in the steam tunnel that Bob Kridle pulled
> allowed them to connect the Cory Hall70, which was the "student"
> system that was were the primary work for 1BSD and 2BSD  occurred
> (ex/csh et al.).   Bob later ran a serial line up the stairwells of
> Cory and picked up the Ingres and the first CAD machine.  Kurt Shoens,
> who was the primary UCB Mail author, hacked the Berknet support into
> his work. One of his primary additions was removing the "delivery"
> part of the mail into a separate program - that he called "delivermail."
>
> Eric Alman was the system administrator of Ing70, so Eric hacked
> delivermail to pass email to and from the Berknet to the ARPANET. All
> was good until Ernie covax showed up and was connected to the UUCP net ;-)
>
> Eric refers to that time as "the email format of the week" and
> sendmail was created to allow him to more easily handle the different
> formats.  By then there was the DARPA 733/822 format (user@host),
> Berknet (host:user), UUCP (host1!host2...!hostn!user) being sources at
> UCB, as well as crap showing up from the IBM Educational System, CSNET
> and various other places trying be exchanged.
>
>
>     The CS department (College of Letters & Science) offices were on
>     the 5th floor.
>
> Ah .. but CS was a division of EECS -- very interesting history about
> that.  The EE Dept was considered top three in the country and #1 by
> many ratings.  They already had a huge history of working with
> industry and had created the Industrial Laison Office (ILO - which was
> how the UNIX BSD tapes were distributed until the creation of CSRG and
> a factoid. The ILO folks advised Fabry on how to run CSRG)).  When CS
> was created, it was forked from the Math Dept. But, the Regents felt
> that if the CS was to have a chance to become a top 10 program, it
> needed to be put under the auspices of the EE Dept like MIT was -- and
> was attached to EE -- hence EECS.
>
>
>     The main campus computer center was in the basement, but they
>     charged by the CPU/second, so only classes used those systems - in
>     my day (1980-1983), the computer center had a CDC 6400 and six or
>     seven DEC PDP-11/70s.
>
> I could be wrong - but ISTR, there was an IBM 360/370 system down
> there also.
>
> The 70s were used to teach CS courses such as the one I taught -- I
> think Bart Miller, Mike Carey and I taught CS-40 a qtr or two after
> Erik was an undergrad.
>
>     BerkNet (RS-232c long haul lines) connected the PDP-11s
>
> Actually, not at all, which was all the more amazing.  There were just
> CAT3 twisted pair with no conditioning.   When we had an electrical
> storm, it was not usually to have the replace the 488/489 transcievers
> in the DHs or DZs.  Kridle got pretty adept and putting machines
> sockets into the boards and he kept a box of chips in his office.
>
>     , and later the VAXen, though Ethernet supplanted it.
>
> It was initially with 3Mbit Xerox boards. We were a little itchy about
> that, so Bob got an optical coupler/repeater (I assume from Xerox, but
> I don't know) in Cory's CAD machine room.   The 10M stuff came about
> two years later, originally 3COM equipment with.  There was a LAN in
> the Ingres machine room, another in the CAD, and another in CSRG's in
> Evans.  The UCB CAD machine was on all three networks as it was in the
> middle.  That is why Sam wrote the original routed stuff -- based on
> things he had seen at PARC for PUP.
>


[-- Attachment #2: Type: text/html, Size: 11015 bytes --]

[TUHS] Re: Berkeley CSRG Building

[arnold]

Clem Cole <clemc@ccc.com> wrote:

> Eric refers to that time as "the email format of the week" and sendmail was
> created to allow him to more easily handle the different formats.  By then
> there was the DARPA 733/822 format (user@host), Berknet (host:user), UUCP
> (host1!host2...!hostn!user) being sources at UCB, as well as crap showing
> up from the IBM Educational System, CSNET and various other places trying
> be exchanged.

Small correction. CSNet didn't happen until the mid-80s, by which time
sendmail was firmly entrenched in the BSD world. Circa 1987/1988, I was
the Unix sysadmin at Emory U., and we got a CSNet connection via Georgia
Tech. It required a leased X.25 (!) line and a special board to put
in one of our Vaxen. There was a driver for it for 4.2BSD but we were
running 4.3. Ron Hutchins at GT and I ported it over to 4.3BSD.

Around that time I did some work with Ease, which was a prettier
language for writing sendmail config files; the sources, including
my mods to it, are in comp.sources.unix somewhere.  I wrote a sendmail
config file *from scratch* using it.  As a result, the Morris worm
passed us by.  :-)

Arnold

P.S. I have *literally* forgotten more about sendmail than most
people ever know. :-)

[TUHS] Re: Berkeley CSRG Building

[arnold]

Luther Johnson <luther.johnson@makerlisp.com> wrote:

> I really enjoy all the history and perspective. A phrase just popped in
> to my head, that I think for me, sums up much of Unix evolution. And it
> is something that not all Linux developers get, as many continue to
> mutate Linux further and further away from traditional Unix ideas:
>
> Necessity is the mother of *convention*.

This was carried even further in Plan 9, making it possible to transparently
network systems of different architectures.

Arnold

[TUHS] Re: Berkeley CSRG Building

[Henry Bent]

[-- Attachment #1: Type: text/plain, Size: 572 bytes --]

On Tue, 13 Aug 2024 at 14:31, <arnold@skeeve.com> wrote:

> Around that time I did some work with Ease, which was a prettier
> language for writing sendmail config files; the sources, including
> my mods to it, are in comp.sources.unix somewhere.  I wrote a sendmail
> config file *from scratch* using it.  As a result, the Morris worm
> passed us by.  :-)
>

I believe the most recent version is preserved at
https://sources.vsta.org/comp.sources.unix/volume25/ease3.5/

There are also older versions in that same archive if for some reason they
are of interest.

-Henry

[-- Attachment #2: Type: text/html, Size: 997 bytes --]

[TUHS] Re: Berkeley CSRG Building

[arnold]

Henry Bent <henry.r.bent@gmail.com> wrote:

> On Tue, 13 Aug 2024 at 14:31, <arnold@skeeve.com> wrote:
>
> > Around that time I did some work with Ease, which was a prettier
> > language for writing sendmail config files; the sources, including
> > my mods to it, are in comp.sources.unix somewhere.  I wrote a sendmail
> > config file *from scratch* using it.  As a result, the Morris worm
> > passed us by.  :-)
> >
>
> I believe the most recent version is preserved at
> https://sources.vsta.org/comp.sources.unix/volume25/ease3.5/

Thanks! I had forgotten that Bruce was the formal maintainer.
But I am named in the README. :-)

Hmmm. I see it was posted on

> Wrapped by vixie@cognition.pa.dec.com on Tue Dec 10 08:45:55 1991

That was well after I'd left Emory.

Thanks!

Arnold

[TUHS] Re: Mailer History -- was Berkeley CSRG Building

[Clem Cole]

[-- Attachment #1: Type: text/plain, Size: 1637 bytes --]

On Tue, Aug 13, 2024 at 2:13 PM <arnold@skeeve.com> wrote:

> ... CSNet didn't happen until the mid-80s, by which time
> sendmail was firmly entrenched in the BSD world. ...


Right -- my bad.  CSNET did not occur until a few years after I had
finished my grad work and left.  ISTR CSNET used MMDF as the mail transport
agent (not sendmail).  But my point was that there were numerous alternate
mail systems, commercial and independent, that had "portals" into and out
of the USENET and the ARPANET and later the Internet - so email headers
were getting warped in strange ways and trying to be able to reliably reply
was often difficult.

Eric was having to deal with some new hacks fairly regularly, and the idea
of a production language to convert format A to B to C was a good one.  The
only real mistake in my mind was that (unlike MMDF) he built the SMTP
daemon in sendmail. MMDF and some of the other systems had programs that
converted to/from a canonical form (usually RFC733/822 style) and then some
sets of external utilities that talked to the "port" - be it UUCP, SMTP,
"PhoneNet", or whatever.   But making it built into sendmail itself, in fact
, was a violation of the UNIX "do one job well" idea and would, of course,
be the attack vector.   The two sad parts of that IMO is that first most
people did not need most of the features of sendmail -- they needed just
SMTP and maybe UUCP, but send sendmail was the SMTPD for BSD, that is what
they had.  But the BBN TCP/IP package that UCB started with had a separate
SMTPD and Eric could have just called it like he did the UUCP subsystem.
ᐧ

[-- Attachment #2: Type: text/html, Size: 4397 bytes --]

[TUHS] Re: Berkeley CSRG Building

[Phil Budne]

arnold@skeeve.com wrote:
> Clem Cole <clemc@ccc.com> wrote:
> > Eric refers to that time as "the email format of the week" and sendmail was
> > created to allow him to more easily handle the different formats.  By then
> > there was the DARPA 733/822 format (user@host), Berknet (host:user), UUCP
> > (host1!host2...!hostn!user) being sources at UCB, as well as crap showing
> > up from the IBM Educational System, CSNET and various other places trying
> > be exchanged.
>
> Small correction. CSNet didn't happen until the mid-80s, by which time
> sendmail was firmly entrenched in the BSD world. Circa 1987/1988, I was
> the Unix sysadmin at Emory U., and we got a CSNet connection via Georgia
> Tech. It required a leased X.25 (!) line and a special board to put
> in one of our Vaxen. There was a driver for it for 4.2BSD but we were
> running 4.3. Ron Hutchins at GT and I ported it over to 4.3BSD.

CSNet had a dialup POTS based service, PhoneNet, using MMDF.  I think
that was Boston University's 4.2BSD VAX 11/780's primary email
connection until we got an Internet connection via Cypress(*) c. 1986,
initially using an 11/725 (a re-packaged 11/730), running, I think,
Ultrix.

Back in the day, global email was a maze of twisty passages with
gateways between the worlds of The ARPANET, BITNET, UUCP, CSNet, etc,
and which required excruciating navigation with percent signs used for
manual routing, tho sendmail configs would route user@host.BITNET etc
to the right gateway with the right incantation.  Often, but not
always, a simple reply would work, but sometimes the rewrites didn't
get the sender address right...

(*) https://docs.lib.purdue.edu/cgi/viewcontent.cgi?article=1565&context=cstech

[TUHS] Re: Mailer History -- was Berkeley CSRG Building

[Dan Cross]

On Tue, Aug 13, 2024 at 3:06 PM Clem Cole <clemc@ccc.com> wrote:
> On Tue, Aug 13, 2024 at 2:13 PM <arnold@skeeve.com> wrote:
>> ... CSNet didn't happen until the mid-80s, by which time
>> sendmail was firmly entrenched in the BSD world. ...
>
> Right -- my bad.  CSNET did not occur until a few years after I had finished my grad work and left.  ISTR CSNET used MMDF as the mail transport agent (not sendmail).  But my point was that there were numerous alternate mail systems, commercial and independent, that had "portals" into and out of the USENET and the ARPANET and later the Internet - so email headers were getting warped in strange ways and trying to be able to reliably reply was often difficult.

I tried to set up MMDF on an RT running AOS once; this was during the
era when Sendmail was getting hacked nearly continuously.  It was kind
of a nifty architecture; different formats used different "channels"
feeding into a common routing core.

It sort of broke down because the inter-message separately wasn't
compatible with any clients we were using; they didn't use the BSD
"mbox" format, but rather, used a set of (4?) `^A`'s as the message
separator.

I'm sure there was some way to change this, but I never investigated
sufficiently to figure out what it was.

> Eric was having to deal with some new hacks fairly regularly, and the idea of a production language to convert format A to B to C was a good one.  The only real mistake in my mind was that (unlike MMDF) he built the SMTP daemon in sendmail. MMDF and some of the other systems had programs that converted to/from a canonical form (usually RFC733/822 style) and then some sets of external utilities that talked to the "port" - be it UUCP, SMTP, "PhoneNet", or whatever.   But making it built into sendmail itself, in fact, was a violation of the UNIX "do one job well" idea and would, of course, be the attack vector.   The two sad parts of that IMO is that first most people did not need most of the features of sendmail -- they needed just SMTP and maybe UUCP, but send sendmail was the SMTPD for BSD, that is what they had.  But the BBN TCP/IP package that UCB started with had a separate SMTPD and Eric could have just called it like he did the UUCP subsystem.
> ᐧ

Zmailer had an architecture kind of like that; at least some folks
were running it locally. But it never really caught on, as far as I
could tell, and eventually qmail and postfix (nee vmailer) provided a
better alternative. I guess some folks still swear by Exim.

I kind of wonder if the days of Unix MTAs as anything other than a
detail are mostly over.

        - Dan C.

[TUHS] Re: Mailer History -- was Berkeley CSRG Building

[George Michaelson]

[-- Attachment #1: Type: text/plain, Size: 3361 bytes --]

Many people lived with mbox as an intermediate delivery form and then did
post fact rewrites on the /var/spool/user file.

I continue to use (n)MH reading from Imap sources with an oauth2 provider
and my mail is in maildir as an intermediate state, very annoying.

I liked mmdf, much because I liked the people who worked on it. Steve Kille
in UCL and Julian Onions in Nottingham amongst others. Marshall rose may
have had a bit of it, and Doug crocker. Smart people. Better coders than me
by far!

G

On Wed, 14 Aug 2024, 16:44 Dan Cross, <crossd@gmail.com> wrote:

> On Tue, Aug 13, 2024 at 3:06 PM Clem Cole <clemc@ccc.com> wrote:
> > On Tue, Aug 13, 2024 at 2:13 PM <arnold@skeeve.com> wrote:
> >> ... CSNet didn't happen until the mid-80s, by which time
> >> sendmail was firmly entrenched in the BSD world. ...
> >
> > Right -- my bad.  CSNET did not occur until a few years after I had
> finished my grad work and left.  ISTR CSNET used MMDF as the mail transport
> agent (not sendmail).  But my point was that there were numerous alternate
> mail systems, commercial and independent, that had "portals" into and out
> of the USENET and the ARPANET and later the Internet - so email headers
> were getting warped in strange ways and trying to be able to reliably reply
> was often difficult.
>
> I tried to set up MMDF on an RT running AOS once; this was during the
> era when Sendmail was getting hacked nearly continuously.  It was kind
> of a nifty architecture; different formats used different "channels"
> feeding into a common routing core.
>
> It sort of broke down because the inter-message separately wasn't
> compatible with any clients we were using; they didn't use the BSD
> "mbox" format, but rather, used a set of (4?) `^A`'s as the message
> separator.
>
> I'm sure there was some way to change this, but I never investigated
> sufficiently to figure out what it was.
>
> > Eric was having to deal with some new hacks fairly regularly, and the
> idea of a production language to convert format A to B to C was a good
> one.  The only real mistake in my mind was that (unlike MMDF) he built the
> SMTP daemon in sendmail. MMDF and some of the other systems had programs
> that converted to/from a canonical form (usually RFC733/822 style) and then
> some sets of external utilities that talked to the "port" - be it UUCP,
> SMTP, "PhoneNet", or whatever.   But making it built into sendmail itself,
> in fact, was a violation of the UNIX "do one job well" idea and would, of
> course, be the attack vector.   The two sad parts of that IMO is that first
> most people did not need most of the features of sendmail -- they needed
> just SMTP and maybe UUCP, but send sendmail was the SMTPD for BSD, that is
> what they had.  But the BBN TCP/IP package that UCB started with had a
> separate SMTPD and Eric could have just called it like he did the UUCP
> subsystem.
> > ᐧ
>
> Zmailer had an architecture kind of like that; at least some folks
> were running it locally. But it never really caught on, as far as I
> could tell, and eventually qmail and postfix (nee vmailer) provided a
> better alternative. I guess some folks still swear by Exim.
>
> I kind of wonder if the days of Unix MTAs as anything other than a
> detail are mostly over.
>
>         - Dan C.
>

[-- Attachment #2: Type: text/html, Size: 4028 bytes --]

[TUHS] Re: Mailer History -- was Berkeley CSRG Building

[Rich Salz]

[-- Attachment #1: Type: text/plain, Size: 539 bytes --]

Several folks at BBN worked on MMDF as well, often as part of the CSNET
project which BBN ran.*  I always thought the five control-A's were a
failure of imagination (never going to send binary?).  MMDF's biggest
failure was synchronous sending -- if you made a typo in an email address
the mail UA would bring it back to edit-mode -- which didn't work for
store-and-forward mail configurations.

Steve Bellovin had a funny quip about CSNET and Usenet, comparing
professors applying for NSF grants and grad students just setting up modems.

[-- Attachment #2: Type: text/html, Size: 625 bytes --]

[TUHS] Re: Mailer History -- was Berkeley CSRG Building

[Clem Cole]

[-- Attachment #1: Type: text/plain, Size: 2452 bytes --]

On Wed, Aug 14, 2024 at 11:44 AM Rich Salz <rich.salz@gmail.com> wrote:

>   I always thought the five control-A's were a failure of imagination
> (never going to send binary?).
>
As always, it's about history and how they got there.  What was reasonable
for one group, over time was not for another.

To be fair, that was Bruce Borden-ism from the Rand Mailer Handler (MH).
Longer before things LIKE Ease/MMDF *et al*.,  Rand was running UNIX
machines on the ARPANET early on (and UUCP did not yet exist).  The mailer
header format needed to be RFC733 in those days.  The use of the "UNIX"
headers - was rather limited to a couple of programs in the Fifth and Sixth
Edition.   Bruce had replaced the Fifth and Sixth edition "mail" program
(which was both the MUI and MTA combined), with separate programs already
as part of creating this new user interface and mailer agent.  Thus, a new
mailbox format was fair game.

It was not a failure of imagination, binary was not an issue for Bruce -
since it was not allowed in RFC733 at that point.   Now, remember that many
UNIX folks on the ARPANET had already switched to MH (that's why we
switched to using it at CMU), so MMDF picking Bruce's format was a good
idea (for them).  Remember, AT&T Research is not on the ARPANET, so the
mail (local to a single machine in those days) could be in any format
they wanted.
RFC733 (and later 822) headers were not an issue, so the funky "From:"
worked fine.

As I pointed out, UCB was not on the ARPANET, so when Kurt wrote the UCB
MUI and split out the MTA as delivermail, he kept the V5/V6 mailbox format (
mbox) from Research.    The issue came when people started using the mail
system as a programmatic messaging scheme (*i.e.,* fork:  some_program |
mail user) and other programs started to parse the output.  More
importantly, getting an ARPANET connection was difficult, so many fewer
sites needed to support the ARPANET (later Internet) header format, whereas
a USENET connection via UUCP was easy.   So >>lots<< of people started to
pick up Mail(1) [Kurt's new MUI) and did not use/know about, much less need
to switch over to MH and RFC733/822.

Thus, the switch from the Research "mbox" format to Borden's scheme started
to become problematic in that if the new utility "knew" about mbox, it had
to be hacked if you were using MH or something that was derived from MH
(like MMDF or PMDF).



ᐧ
ᐧ

[-- Attachment #2: Type: text/html, Size: 5334 bytes --]

[TUHS] Re: Mailer History -- was Berkeley CSRG Building

[Steffen Nurpmeso]

Clem Cole wrote in
 <CAC20D2Nin-yvELys6QHGZjw1otuLegXy3keyg4-E1kd3ZME+KA@mail.gmail.com>:
 |On Wed, Aug 14, 2024 at 11:44 AM Rich Salz <rich.salz@gmail.com> wrote:
 |
 |>   I always thought the five control-A's were a failure of imagination
 |> (never going to send binary?).
 |>
 |As always, it's about history and how they got there.  What was reasonable
 |for one group, over time was not for another.
 |
 |To be fair, that was Bruce Borden-ism from the Rand Mailer Handler (MH).
 |Longer before things LIKE Ease/MMDF *et al*.,  Rand was running UNIX
 |machines on the ARPANET early on (and UUCP did not yet exist).  The mailer
 |header format needed to be RFC733 in those days.  The use of the "UNIX"
 |headers - was rather limited to a couple of programs in the Fifth and Sixth
 |Edition.   Bruce had replaced the Fifth and Sixth edition "mail" program
 |(which was both the MUI and MTA combined), with separate programs already
 |as part of creating this new user interface and mailer agent.  Thus, a new
 |mailbox format was fair game.
 |
 |It was not a failure of imagination, binary was not an issue for Bruce -
 |since it was not allowed in RFC733 at that point.   Now, remember that many

Depending on what you mean by "at that point" i think here you
misremember.  (To the contrary 733 and also 822 allow practically
anything, it was RFC 2822 from 2001 which then at least says "The
NUL character (ASCII value 0) was once allowed, but is no longer
for compatibility reasons".  Except for NUL (comments) can still
as of today contain the now-called obs-NO-WS-CTL defined as
 obs-NO-WS-CTL   = %d1-8 / %d11 / %d12 / %d14-31 / %d127;
  ; US-ASCII controls except CR, LF, and whitespace)

 |UNIX folks on the ARPANET had already switched to MH (that's why we
 |switched to using it at CMU), so MMDF picking Bruce's format was a good
 |idea (for them).  Remember, AT&T Research is not on the ARPANET, so the
 |mail (local to a single machine in those days) could be in any format
 |they wanted.
 |RFC733 (and later 822) headers were not an issue, so the funky "From:"
 |worked fine.
 |
 |As I pointed out, UCB was not on the ARPANET, so when Kurt wrote the UCB
 |MUI and split out the MTA as delivermail, he kept the V5/V6 mailbox \
 |format (
 |mbox) from Research.    The issue came when people started using the mail
 |system as a programmatic messaging scheme (*i.e.,* fork:  some_program |
 |mail user) and other programs started to parse the output.  More
 |importantly, getting an ARPANET connection was difficult, so many fewer
 |sites needed to support the ARPANET (later Internet) header format, whereas
 |a USENET connection via UUCP was easy.   So >>lots<< of people started to
 |pick up Mail(1) [Kurt's new MUI) and did not use/know about, much less need
 |to switch over to MH and RFC733/822.
 |
 |Thus, the switch from the Research "mbox" format to Borden's scheme started
 |to become problematic in that if the new utility "knew" about mbox, it had
 |to be hacked if you were using MH or something that was derived from MH
 |(like MMDF or PMDF).

As a remark, the MBOX format as standardized by POSIX decades ago

  line beginning with From<space>
  [one or more header fields; see Commands in mailx (on page 3112)]
  empty line
  [zero or more body lines
  empty line]
  [line beginning with From<space>...]

is pretty clear (the BSD Mail i took maintainership of only did it
too laxe, and it took prodding by Dr. Werner Fink of SuSE for me
to realize *how* much, in 2019).  And even more so the MBOX format
as standardized by RFC 4155!  Compare to the Unicode people, who
advise on looking for and then fully trust the BOM (U+FEFF), which
is only 16-bit.

(And also, with MIME, which was standardized in its iterated, current
version 28 years ago, and RFC 4155 from 2005, one would normally
save content-encoded data in the thus unambiguous,
non-misinterpretable database that MBOX is; the IETF has not yet
iterated RFC 4155, so all the "UTF-8" email extensions do not
apply, and 4155 states: "Eight-bit data within the stream MUST be
converted to a seven-bit form (using appropriate, standardized
encoding) and appropriately tagged (with the correct header
fields) before the database is transferred".)

--steffen
|
|Der Kragenbaer,                The moon bear,
|der holt sich munter           he cheerfully and one by one
|einen nach dem anderen runter  wa.ks himself off
|(By Robert Gernhardt)

[TUHS] Re: Mailer History -- was Berkeley CSRG Building

[Rich Salz]

[-- Attachment #1: Type: text/plain, Size: 396 bytes --]

>
>   I always thought the five control-A's were a failure of imagination
>> (never going to send binary?).
>>
> As always, it's about history and how they got there.  What was reasonable
> for one group, over time was not for another.
>

True.  I joined BBN after the TCP/IP flag day, after 4.2BSD, and so on.
When I used MH, it kept one message/file; I only know the control-a thing
from MMDF.

[-- Attachment #2: Type: text/html, Size: 949 bytes --]

[TUHS] Re: Mailer History -- was Berkeley CSRG Building

[Clem Cole]

[-- Attachment #1: Type: text/plain, Size: 2793 bytes --]

On Wed, Aug 14, 2024 at 2:18 PM Steffen Nurpmeso <steffen@sdaoden.eu> wrote:

> Depending on what you mean by "at that point" i think here you
> misremember.  (To the contrary 733 and also 822 allow practically anything
>
Page 3 of SMTPD -- RFC 821 [which the 822 sits on top of says]:

"Commands and Replies are composed of characters from the ASCII character
set.  When a transport service provides an 8-bit byte (octet) transmission
channel, each 7-bit character  is transmitted right justified in an octet
with the *high order bit cleared to zero*"


BTW, this was specified in 821 because it had been a factor in earlier
experience of the ARPANET [733 and using FTP as a mail transport, which was
how much of this all started], and binary was explicitly not allowed.   822
force 7-bit ASCII because of the earlier issues of things like CDC's
display code (what a nightmare), much less EBCDIC.  The key was that those
of us in the ARPANET community could not allow "anything," - but we did
have to detail what was there.

Some of us lived in this world and wrote programs that dealt with those
constraints at the time.  I am relaying to you what it was and how it
happened.  As I said, we have what we have because that was what we were
living with -- two distinct worlds, the ARPANET community - which was
setting the standards for interchange, and UNIX (USENET), which was de
facto and growing because it cost little to join it.


> As a remark, the MBOX format as standardized by POSIX decades ago
>
Ouch -- I was part of POSIX and /usr/group before that. And had >>nothing<<
to do with it.   The POSIX definition was at least 15 years  After Bruce
wrote MH and Kurt did delivermail and if I count I >>suspect<< it is
correctly closer to 20-25.  mbox was created I believe by Ken (Doug do you
remember?), but I'm not sure who actually wrote the original "mail" program
for Fifth edition (maybe Fourth) - which as I said was both an MUI and a
MTA.   But that development was over 15 yrs before we started the
/usr/group standard, much less the POSIX ones.

You are probably correct that until it was formally specified in the POSIX
definition, the format was defined originally in Ken's code, then Kurt's
and finally in Eric's.   IIRC, Bruce actually had a man page in MH that
described his format that used the ^A characters, but I would have to
rummage through old sources to be 100% certain.  Certainly, later
distributions of it did describe it, and MMDF may have also - but I'm not
sure.

BTW: by about 4.2BSD time (maybe a little earlier), particularly because of
sendmail - the MH system (which had left Rand and was then being supported
by someone else ??UC Irvine maybe??) had been hacked to handle the mbox
format


ᐧ

[-- Attachment #2: Type: text/html, Size: 5391 bytes --]

[TUHS] Re: Mailer History -- was Berkeley CSRG Building

[Clem Cole]

[-- Attachment #1: Type: text/plain, Size: 538 bytes --]

On Wed, Aug 14, 2024 at 2:47 PM Rich Salz <rich.salz@gmail.com> wrote:

>   I always thought the five control-A's were a failure of imagination
>>> (never going to send binary?).
>>>
>> As always, it's about history and how they got there.  What was
>> reasonable for one group, over time was not for another.
>>
>
> True.  I joined BBN after the TCP/IP flag day, after 4.2BSD, and so on.
> When I used MH, it kept one message/file; I only know the control-a thing
> from MMDF.
>
Rich, that was a much later version.
ᐧ

[-- Attachment #2: Type: text/html, Size: 1867 bytes --]

[TUHS] Re: Mailer History -- was Berkeley CSRG Building

[Clem Cole]

[-- Attachment #1: Type: text/plain, Size: 3024 bytes --]

s/to detail/to deal with/ - I hate autocorrect
ᐧ

On Wed, Aug 14, 2024 at 2:50 PM Clem Cole <clemc@ccc.com> wrote:

>
> On Wed, Aug 14, 2024 at 2:18 PM Steffen Nurpmeso <steffen@sdaoden.eu>
> wrote:
>
>> Depending on what you mean by "at that point" i think here you
>> misremember.  (To the contrary 733 and also 822 allow practically
>> anything
>>
> Page 3 of SMTPD -- RFC 821 [which the 822 sits on top of says]:
>
> "Commands and Replies are composed of characters from the ASCII character
> set.  When a transport service provides an 8-bit byte (octet) transmission
> channel, each 7-bit character  is transmitted right justified in an octet
> with the *high order bit cleared to zero*"
>
>
> BTW, this was specified in 821 because it had been a factor in earlier
> experience of the ARPANET [733 and using FTP as a mail transport, which was
> how much of this all started], and binary was explicitly not allowed.   822
> force 7-bit ASCII because of the earlier issues of things like CDC's
> display code (what a nightmare), much less EBCDIC.  The key was that those
> of us in the ARPANET community could not allow "anything," - but we did
> have to detail what was there.
>
> Some of us lived in this world and wrote programs that dealt with those
> constraints at the time.  I am relaying to you what it was and how it
> happened.  As I said, we have what we have because that was what we were
> living with -- two distinct worlds, the ARPANET community - which was
> setting the standards for interchange, and UNIX (USENET), which was de
> facto and growing because it cost little to join it.
>
>
>> As a remark, the MBOX format as standardized by POSIX decades ago
>>
> Ouch -- I was part of POSIX and /usr/group before that. And had
> >>nothing<< to do with it.   The POSIX definition was at least 15 years
> After Bruce wrote MH and Kurt did delivermail and if I count I >>suspect<<
> it is correctly closer to 20-25.  mbox was created I believe by Ken (Doug
> do you remember?), but I'm not sure who actually wrote the original "mail"
> program for Fifth edition (maybe Fourth) - which as I said was both an MUI
> and a MTA.   But that development was over 15 yrs before we started the
> /usr/group standard, much less the POSIX ones.
>
> You are probably correct that until it was formally specified in the POSIX
> definition, the format was defined originally in Ken's code, then Kurt's
> and finally in Eric's.   IIRC, Bruce actually had a man page in MH that
> described his format that used the ^A characters, but I would have to
> rummage through old sources to be 100% certain.  Certainly, later
> distributions of it did describe it, and MMDF may have also - but I'm not
> sure.
>
> BTW: by about 4.2BSD time (maybe a little earlier), particularly because
> of sendmail - the MH system (which had left Rand and was then being
> supported by someone else ??UC Irvine maybe??) had been hacked to handle
> the mbox format
>
>
> ᐧ
>

[-- Attachment #2: Type: text/html, Size: 6178 bytes --]

[TUHS] Re: Mailer History -- was Berkeley CSRG Building

[Clem Cole]

[-- Attachment #1: Type: text/plain, Size: 892 bytes --]

On Wed, Aug 14, 2024 at 2:51 PM Clem Cole <clemc@ccc.com> wrote:

>
>
> On Wed, Aug 14, 2024 at 2:47 PM Rich Salz <rich.salz@gmail.com> wrote:
>
>>   I always thought the five control-A's were a failure of imagination
>>>> (never going to send binary?).
>>>>
>>> As always, it's about history and how they got there.  What was
>>> reasonable for one group, over time was not for another.
>>>
>>
>> True.  I joined BBN after the TCP/IP flag day, after 4.2BSD, and so on.
>> When I used MH, it kept one message/file; I only know the control-a thing
>> from MMDF.
>>
> Rich, that was a much later version.
>
FWIW: The big complaint about MH was not the mailbox format, it was that
Bruce stored your files in separate files, which on a small machine like a
PDP-11, eats up inode space, although it made doing things like grep and
finding something easy.

ᐧ
>
ᐧ

[-- Attachment #2: Type: text/html, Size: 3295 bytes --]

[TUHS] Re: Mailer History -- was Berkeley CSRG Building

[segaloco via TUHS]

On Wednesday, August 14th, 2024 at 9:45 AM, Clem Cole <clemc@ccc.com> wrote:

> 
> ...
> The issue came when people started using the mail system as a programmatic messaging scheme (i.e., fork: some_program | mail user) and other programs started to parse the output.
> ...

Mail as IPC...that's what I'm reading from that anyway...now that's an interesting concept.  Did that idea ever grow any significant legs?  I can't tell if the general concept is clever or systems abuse, in those days it seems like it could've gone either way.

I guess it sorta did survive in the form of automated systems today expecting specially formatted emails to trigger "stuff" to happen.

- Matt G.

[TUHS] Re: Mailer History -- was Berkeley CSRG Building

[Clem Cole]

[-- Attachment #1: Type: text/plain, Size: 2968 bytes --]

Matt - I'm going to BCC: TUHS and move this to COFF - since while UNIX was
certainly in the mix in all this, it was hardly first or the only place it
happenned.

On Wed, Aug 14, 2024 at 2:59 PM segaloco via TUHS <tuhs@tuhs.org> wrote:

> On Wednesday, August 14th, 2024 at 9:45 AM, Clem Cole <clemc@ccc.com>
> wrote:
>
> >
> > ...
> > The issue came when people started using the mail system as a
> programmatic messaging scheme (i.e., fork: some_program | mail user) and
> other programs started to parse the output.
> > ...
>
> Mail as IPC...that's what I'm reading from that anyway...now that's an
> interesting concept.

It's kind of funny the history.  ARPANET gives us FTP as a way to
exchange files.   So, people figure out how to hack the mailer to call FTP
to send a file remotely and set up a submit a cron/batch submission, a.k.a
RJE.  This is encouraged by DARPA because part of the justification of the
ARAPNET was to be able to share resources, and this enables supercomputers
of the day to be able to provide cycles to DARPA folks who might not have
access to larger systems.   Also, remember, mailers were local to systems
at that point.


So someone gets the bright idea to hooker the mailer into this system --
copy the "mail file" and set up a remote job to mail it locally.  Let's
just say this prioves to be a cool idea and the idea of intersystem email
begins in the >>ARPANET<< community.

So the idea of taking it to the next level was not that far off.  The
mailer transports started to offer (limited) features to access services.
By the time of Kurt's "delivermail" but he added a feature, thinking it was
system logs that allowed specific programs to be called.   In fact, it may
have been invented elsewhere but before Eric would formalize "vacation" -
Jim Kleckner and I hacked together a "awk" script to do that function on
the UCB CAD 4.1 systems.  I showed it to Sam and a few other people, and I
know it went from Cory to Evans fairly quickly.   Vacation(1) was written
shortly there after to be a bit more flexible than our original script.



  Did that idea ever grow any significant legs?

I guess the word here is significant.  It certainly was used where it made
sense.  In the CAD group, we had simulations that might run for a few
days.   We used to call the mailer every so often to send status and
sometimes do something like a checkpoint.  It lead to Sam writing syslogd,
particularly after Joy created UNIX domain sockets.   But I can say we used
it a number of places in systems oriented or long running code before
syslogd as a scheme to log errors, deal with stuff.



>   I can't tell if the general concept is clever or systems abuse, in those
> days it seems like it could've gone either way.
>
> I guess it sorta did survive in the form of automated systems today
> expecting specially formatted emails to trigger "stuff" to happen.

Exactly.
ᐧ

[-- Attachment #2: Type: text/html, Size: 5715 bytes --]

[TUHS] Re: Mailer History -- was Berkeley CSRG Building

[Kevin Bowling]

On Wed, Aug 14, 2024 at 11:59 AM segaloco via TUHS <tuhs@tuhs.org> wrote:
>
> On Wednesday, August 14th, 2024 at 9:45 AM, Clem Cole <clemc@ccc.com> wrote:
>
> >
> > ...
> > The issue came when people started using the mail system as a programmatic messaging scheme (i.e., fork: some_program | mail user) and other programs started to parse the output.
> > ...
>
> Mail as IPC...that's what I'm reading from that anyway...now that's an interesting concept.  Did that idea ever grow any significant legs?  I can't tell if the general concept is clever or systems abuse, in those days it seems like it could've gone either way.

I like Clem's answer on mail IPC/RPC.

To add I have heard some stories of NNTP being used once upon a time
at some service providers the way ansible/mcollective/salt might be
used to orchestrate UNIX host configurations and application
deployments.  The concept of Control messages is somewhat critical to
operations, so it's not totally crazy, but isolating article flows
would give me some heartburn if the thing has privileged system
access.. would probably want it on a totally distinct
instance+port+configuration.

Email and Usenet both have some nice properties of implementing a
"Message Queue" like handling offline hosts when they come back.  But
the complexity of mail and nntp implementations lean more towards
system abuse IMO.

> I guess it sorta did survive in the form of automated systems today expecting specially formatted emails to trigger "stuff" to happen.
>
> - Matt G.

[TUHS] Re: Mailer History -- was Berkeley CSRG Building

[Steffen Nurpmeso]

Clem Cole wrote in
 <CAC20D2O33LZvF2-gn2J8yB_84jk8jfPAdgsvyT476qv54o9U1w@mail.gmail.com>:
 |On Wed, Aug 14, 2024 at 2:18 PM Steffen Nurpmeso <steffen@sdaoden.eu> \
 |wrote:
 |> Depending on what you mean by "at that point" i think here you
 |> misremember.  (To the contrary 733 and also 822 allow practically \
 |> anything
 |>
 |Page 3 of SMTPD -- RFC 821 [which the 822 sits on top of says]:
 |
 |"Commands and Replies are composed of characters from the ASCII character
 |set.  When a transport service provides an 8-bit byte (octet) transmission
 |channel, each 7-bit character  is transmitted right justified in an octet
 |with the *high order bit cleared to zero*"

Sure, that is the 8-bit thing that i introduced into the
discussion.  You all were talking about binary, which means
"lots of non-printable characters".

 |BTW, this was specified in 821 because it had been a factor in earlier
 |experience of the ARPANET [733 and using FTP as a mail transport, which was
 |how much of this all started], and binary was explicitly not allowed.   822
 |force 7-bit ASCII because of the earlier issues of things like CDC's
 |display code (what a nightmare), much less EBCDIC.  The key was that those
 |of us in the ARPANET community could not allow "anything," - but we did
 |have to detail what was there.

Ok, mails had to be transferred somehow.  So without binary
transport no binary storage.  (RFC 822 allows any ASCII character,
except that CRLF ends a line.)

 |Some of us lived in this world and wrote programs that dealt with those
 |constraints at the time.  I am relaying to you what it was and how it
 |happened.  As I said, we have what we have because that was what we were
 |living with -- two distinct worlds, the ARPANET community - which was
 |setting the standards for interchange, and UNIX (USENET), which was de
 |facto and growing because it cost little to join it.
 |
 |> As a remark, the MBOX format as standardized by POSIX decades ago
 |>
 |Ouch -- I was part of POSIX and /usr/group before that. And had >>nothing<<
 |to do with it.   The POSIX definition was at least 15 years  After Bruce
 |wrote MH and Kurt did delivermail and if I count I >>suspect<< it is
 |correctly closer to 20-25.  mbox was created I believe by Ken (Doug do you
 |remember?), but I'm not sure who actually wrote the original "mail" program
 |for Fifth edition (maybe Fourth) - which as I said was both an MUI and a
 |MTA.   But that development was over 15 yrs before we started the
 |/usr/group standard, much less the POSIX ones.

That all is very interesting.  For now i always presumed s2/mail.c
was mostly written by Ken Thompson?  Yes, Research V5 already did
some sort of From_, and ensured a blank line when doing concat()
onto the target users "MBOX" file.

 |You are probably correct that until it was formally specified in the POSIX
 |definition, the format was defined originally in Ken's code, then Kurt's
 |and finally in Eric's.   IIRC, Bruce actually had a man page in MH that
 |described his format that used the ^A characters, but I would have to
 |rummage through old sources to be 100% certain.  Certainly, later
 |distributions of it did describe it, and MMDF may have also - but I'm not
 |sure.

The difference, and what i meant, is that POSIX says "one or more
header lines".  Compare this to Kurt Shoen's ishead() from 1978,
where he only tests a single line by itself.  (But "From " plus
date plus user of maximally 17 bytes is still better than five
times ^A i would think, except that consecutive ^A are rare.)

So i would think the problem stems from the fact that in early
[mM]ails there was no separating newline in between the "From "
line and message text *unless* there was some header to put:

  fprintf(fout, "From %s %s", myname, date);
  puthead(hp, fout);
  while ((c = getc(fo)) != EOF)
    putc(c, fout);

  ->

  puthead(hp, fo)
         struct header *hp;
         FILE *fo;
  {
         if (hp->h_to != NOSTR)
                 fprintf(fo, "To: %s\n", hp->h_to);
         if (hp->h_subj != NOSTR)
                 fprintf(fo, "Subj: %s\n", hp->h_subj);
         if (hp->h_cc != NOSTR)
                 fprintf(fo, "Cc: %s\n", hp->h_cc);
         if (hp->h_to != NOSTR || hp->h_subj != NOSTR || hp->h_cc != NOSTR)
                 putc('\n', fo);
         return(0);
  }

So there *could* be no separating empty line after the
(so-called) From_ line.  This actually changed with POSIX, and it
took more than fourty years until Werner Fink pointed to the fact
that this "one ore more header lines" can be taken into account
when doing From_ line detection in a MBOX file.
Or 31 years, if it was POSIX 1988 which brought that rule.

For a later born one

- it is de facto hard to understand why noone cared for some sort
  of content encoding for neither of SMTP nor text messages.

  I mean, if one reads the early RFCs, before 1975, say, it was
  all unbelievable "direct", trial and error, discovery, etc.
  But, take for example RFC 698 from July 1975, "TELNET EXTENDED
  ASCII OPTION", which says

    Several sites[.] for example MIT-AI, use keyboards which use
    almost all 128 characters as printable characters, and use one
    or more additional bits as "control' bits as command modifiers
    [.] several characters cannot be entered as text because they
    are used for control purposes, such as the greek letter "beta'
    which on a TELNET connection is CONTROL-C and is used for
    stopping ones job.

  Ie "control by external player" was not only the usual thing,
  but it was recognized as causing problems.

  It is fascinating to read Postel's RFC 767 from 1980, "A
  Structured Format for Transmission of Multi-Media Documents".

  To the contrary it took until the early 90s until it was no
  longer expected that receivers "know" (like RFC 767, which
  assumes users know how to interpret data formats to the odd
  bit) by converting data to harmless text "garbage" which needs
  dedicated and declared support on the receiver side to become
  interpreted.

  And i hope IETF's new thing SML will not bring back such.

- Why the 8-bit problem, at all?  Already RFC 354 says:
  The transfer byte size must be 8 bits.

It is really amazing that in hindsight to data formats like RFC
767 encoding for (the) local storage (that) MBOX (is) was not
worth an RFC.  (Until MIME came which *i* use to circumvent any
possible problem; it must be said that especially in the OSS world
people *explicitly* do not do this, so that messages with >From
quoting, aka *without* MIME, can still be seen.)

 |BTW: by about 4.2BSD time (maybe a little earlier), particularly because of
 |sendmail - the MH system (which had left Rand and was then being supported
 |by someone else ??UC Irvine maybe??) had been hacked to handle the mbox
 |format

These days anyone runs away from MBOX.  Even the dovecot IMAP(++)
server no longer uses it (by default) i think, since not too long
ago.  I think they all went the Maildir way, which also stores
one message per file.  I love MBOX (with MIME encoded messages
etc), however i -- for the MUA i maintain -- still have the way to
go they all have passed, and that is external index files etc.
Netscape's mailer had that thirty years ago...

--steffen
|
|Der Kragenbaer,                The moon bear,
|der holt sich munter           he cheerfully and one by one
|einen nach dem anderen runter  wa.ks himself off
|(By Robert Gernhardt)

[TUHS] Re: Berkeley CSRG Building

[Eric Allman via TUHS]

[-- Attachment #1: Type: text/plain, Size: 2367 bytes --]

Clem, normally your history is completely accurate, but this time I need 
to provide a few corrections.

On 2024-08-12 10:34, Clem Cole wrote:
> csh et al.).   Bob later ran a serial line up the stairwells of Cory 
> and picked up the Ingres and the first CAD machine.  Kurt Shoens, who 
> was the primary UCB Mail author, hacked the Berknet support into his 
> work.  One of his primary additions was removing the "delivery" part 
> of the mail into a separate program - that he called "delivermail."
Actually, I'm the one who separated the delivery into another program 
that was indeed called "delivermail". I modified /bin/mail to do actual 
delivery if it was invoked with -d, otherwise call delivermail. I took 
out the hacks for networks such as UUCP and moved them to delivermail.
>
> Eric Alman was the system administrator of Ing70, so Eric hacked 
> delivermail to pass email to and from the Berknet to the ARPANET. All 
> was good until Ernie covax showed up and was connected to the UUCP net ;-)
That part is correct. In fact, the ARPANET was the reason for 
delivermail in the first place. The BerkNet hacks had already been done 
to /bin/mail, I think by Eric Schmidt.
>
> Eric refers to that time as "the email format of the week" and 
> sendmail was created to allow him to more easily handle the different 
> formats.  By then there was the DARPA 733/822 format (user@host), 
> Berknet (host:user), UUCP (host1!host2...!hostn!user) being sources at 
> UCB, as well as crap showing up from the IBM Educational System, CSNET 
> and various other places trying be exchanged.

Don't forget DECnet (host::user) and things like some mercifully dead UK 
network that reversed the domain names, so this mailing list would be 
"tuhs@org.tuhs". And the compiled in configuration that delivermail used 
was becoming unwieldy as the world get bigger, hence a configuration file.

But the real reason for sendmail was the advent of the Internet, which 
used an entirely new mail protocol called SMTP. That required 
implementing queuing and better error reporting than returning an exit 
status. Bill Joy talked me into it because "you know mail better than 
anyone around here." I didn't expect it to turn into the cornerstone of 
my professional life. There are more stories there, but they are off topic.

eric


[-- Attachment #2: Type: text/html, Size: 4071 bytes --]

[TUHS] Re: Berkeley CSRG Building

[Clem Cole]

[-- Attachment #1: Type: text/plain, Size: 2518 bytes --]

thank you.
ᐧ

On Fri, Aug 16, 2024 at 5:08 PM Eric Allman <tuhs@eric.allman.name> wrote:

> Clem, normally your history is completely accurate, but this time I need
> to provide a few corrections.
> On 2024-08-12 10:34, Clem Cole wrote:
>
> csh et al.).   Bob later ran a serial line up the stairwells of Cory and
> picked up the Ingres and the first CAD machine.  Kurt Shoens, who was the
> primary UCB Mail author, hacked the Berknet support into his work.  One of
> his primary additions was removing the "delivery" part of the mail into a
> separate program - that he called "delivermail."
>
> Actually, I'm the one who separated the delivery into another program that
> was indeed called "delivermail". I modified /bin/mail to do actual delivery
> if it was invoked with -d, otherwise call delivermail. I took out the hacks
> for networks such as UUCP and moved them to delivermail.
>
>
> Eric Alman was the system administrator of Ing70, so Eric hacked
> delivermail to pass email to and from the Berknet to the ARPANET.  All
> was good until Ernie covax showed up and was connected to the UUCP net ;-)
>
> That part is correct. In fact, the ARPANET was the reason for delivermail
> in the first place. The BerkNet hacks had already been done to /bin/mail, I
> think by Eric Schmidt.
>
>
> Eric refers to that time as "the email format of the week" and sendmail
> was created to allow him to more easily handle the different formats.  By
> then there was the DARPA 733/822 format (user@host), Berknet (host:user),
> UUCP (host1!host2...!hostn!user) being sources at UCB, as well as crap
> showing up from the IBM Educational System, CSNET and various other places
> trying be exchanged.
>
> Don't forget DECnet (host::user) and things like some mercifully dead UK
> network that reversed the domain names, so this mailing list would be
> "tuhs@org.tuhs" <tuhs@org.tuhs>. And the compiled in configuration that
> delivermail used was becoming unwieldy as the world get bigger, hence a
> configuration file.
>
> But the real reason for sendmail was the advent of the Internet, which
> used an entirely new mail protocol called SMTP. That required implementing
> queuing and better error reporting than returning an exit status. Bill Joy
> talked me into it because "you know mail better than anyone around here." I
> didn't expect it to turn into the cornerstone of my professional life.
> There are more stories there, but they are off topic.
>
> eric
>
>

[-- Attachment #2: Type: text/html, Size: 4406 bytes --]

[TUHS] Re: Mailer History -- was Berkeley CSRG Building

[Eric Allman via TUHS]

[-- Attachment #1: Type: text/plain, Size: 876 bytes --]


On 2024-08-14 12:20, Clem Cole wrote:
>
>     Did that idea ever grow any significant legs?
>
> I guess the word here is significant.  It certainly was used where it 
> made sense.  In the CAD group, we had simulations that might run for a 
> few days.   We used to call the mailer every so often to send status 
> and sometimes do something like a checkpoint. It lead to Sam writing 
> syslogd, particularly after Joy created UNIX domain sockets.   But I 
> can say we used it a number of places in systems oriented or long 
> running code before syslogd as a scheme to log errors, deal with stuff.

Another correction: I wrote syslogd as part of the delivermail/sendmail 
project, but I did intentionally make it generic so it could be used by 
other services. Interestingly, I even got it to work on v6 using 
something called "mpx files".

eric

[-- Attachment #2: Type: text/html, Size: 1758 bytes --]

[TUHS] Re: Berkeley CSRG Building

[George Michaelson]

[-- Attachment #1: Type: text/plain, Size: 1667 bytes --]

On Sat, 17 Aug 2024, 7:08 am Eric Allman via TUHS, <tuhs@tuhs.org> wrote:

> Don't forget DECnet (host::user) and things like some mercifully dead UK
> network that reversed the domain names, so this mailing list would be
> "tuhs@org.tuhs" <tuhs@org.tuhs>. And the compiled in configuration that
> delivermail used was becoming unwieldy as the world get bigger, hence a
> configuration file.
>
> eric
>

Steve Kille famously said on the JANET (for that was the network) uk mail
list for mail of the UK.ac decision "its research and its ok to experiment"
- the main advantage was clarity of the scoping to which element to look at
going to far off faroffia:  it was the rightmost element in the token list
for you normal people and the leftmost for us. Since we are western
alphabet and alrwsdy parsing the user@host left to right it meant in
principle the channel for faroffia was found faster from a shorter index
token starting from 0.

Jim at, Leeds uni and then heriot-watt wrote the sendmail.cf to
dis-un-combobulate uk.ac to ac.uk which obviously many many sysadmins in
the UK ran with. He was really meant to be doing functional programming
research I think. We shared an office for a few months at Leeds, it was ex
English school and reputedly where Tolkien sat out his days before
Cambridge came good. Leeds could have taken free(ish) mail from York on x25
and preferred to dial the Heriot-Watt in edinburgh to get uucp. Acoustic
coupler modem days. I think has they known, Charles Forsyth in York would
have done uucp over Janet /x25 but people sometimes do the other thing, not
because it's hard but just because.

G

>

[-- Attachment #2: Type: text/html, Size: 2462 bytes --]

[TUHS] Re: Berkeley CSRG Building

[Eric Allman via TUHS]

[-- Attachment #1: Type: text/plain, Size: 2879 bytes --]


On 2024-08-16 17:47, George Michaelson wrote:
>
>
> On Sat, 17 Aug 2024, 7:08 am Eric Allman via TUHS, <tuhs@tuhs.org> wrote:
>
>     Don't forget DECnet (host::user) and things like some mercifully
>     dead UK network that reversed the domain names, so this mailing
>     list would be "tuhs@org.tuhs" <mailto:tuhs@org.tuhs>. And the
>     compiled in configuration that delivermail used was becoming
>     unwieldy as the world get bigger, hence a configuration file.
>
>     eric
>
>
> Steve Kille famously said on the JANET (for that was the network) uk 
> mail list for mail of the UK.ac decision "its research and its ok to 
> experiment" - the main advantage was clarity of the scoping to which 
> element to look at going to far off faroffia:  it was the rightmost 
> element in the token list for you normal people and the leftmost for 
> us. Since we are western alphabet and alrwsdy parsing the user@host 
> left to right it meant in principle the channel for faroffia was found 
> faster from a shorter index token starting from 0.

Ah yes, I was trying to remember JANET but was too lazy to do the research.

Honestly, I thought that JANET got it right and the rest of us 
different, so that:

user@top.middle.bottom (e.g., eric@edu.berkeley.cs)

would allow strict left-to-right parsing. Actually I wanted 
cs.berkeley.edu:eric — if that was true everywhere, sendmail would have 
been so much easier. A major reason for very generic rewriting rules is 
that basic parsing algorithms (notably LALR(1)) couldn't be made 
generic. At Berkeley

uunet!foo!bar@berkeley.edu

meant that the message should be sent to the UUCP host (ucbvax at the 
time, iirc, which from Ing70 translated to "ucbvax:uunet!foo!bar"), but

host::user@decwrl.com

should be sent to decwrl unchanged. See the book "!%@:: A directory of 
Electronic Mail Addressing & Networks" for a taste of just how bad it was.

>
> Jim at, Leeds uni and then heriot-watt wrote the sendmail.cf 
> <http://sendmail.cf> to dis-un-combobulate uk.ac <http://uk.ac> to 
> ac.uk <http://ac.uk> which obviously many many sysadmins in the UK ran 
> with. He was really meant to be doing functional programming research 
> I think. We shared an office for a few months at Leeds, it was ex 
> English school and reputedly where Tolkien sat out his days before 
> Cambridge came good. Leeds could have taken free(ish) mail from York 
> on x25 and preferred to dial the Heriot-Watt in edinburgh to get uucp. 
> Acoustic coupler modem days. I think has they known, Charles Forsyth 
> in York would have done uucp over Janet /x25 but people sometimes do 
> the other thing, not because it's hard but just because.
>
> G

Heh. It reminds me of some of Teus Hagen's escapades in the early days. 
But Teus was earlier, when networking meant UUCP.

eric

[-- Attachment #2: Type: text/html, Size: 5178 bytes --]

[TUHS] Re: Berkeley CSRG Building

[Jaap Akkerhuis via TUHS]

[-- Attachment #1: Type: text/plain, Size: 584 bytes --]

Hi Eric,

> On 17 Aug 2024, at 03:20, Eric Allman via TUHS <tuhs@tuhs.org> wrote:
> 
> Heh. It reminds me of some of Teus Hagen's escapades in the early days. But Teus was earlier, when networking meant UUCP.
> 

Yes.  I do remember him writing the first sendmail.cf file.  Many people
used his version which you could always recognize because of the comment
"Will this ever work?” or something like that.  I do remember I 
found it back in various Ultix and other distributions.  This is likely
because it ended up on the EUUG networking distribution tapes.

	jaap


[-- Attachment #2: Type: text/html, Size: 1405 bytes --]

[TUHS] Re: Berkeley CSRG Building

[Jon Forrest]

On 8/16/24 11:25 AM, Noel Chiappa wrote:
>      > From: Larry McVoy
> 
> {Moving this to COFF, as it's not UNIX-related. I'll have another reply there
> as well, about the social media point.}
> 
>      > The amazing thing, to me, is I was a CS student in very early 1980's
>      > and I had no idea of the history behind the arpanet.
> 
> I don't think that was that uncommon; at MIT (slightly earlier, I think -
> -'74-'77 for me) the undergrad's weren't learning anything about networking
> there either, then.


I was at UC Santa Barbara as an undergrad from 1973 to 1976, and then
back from 1978 to 1985. UCSB was one of the original 4 Arpanet nodes
but virtually nobody outside of a small group in EE knew about it.
I think the Culler-Harrison time sharing research and the speech
research at UCSB and SCRL (Speed Communication Research Laboratory,
a private speech research lab where many of the people were also
affiliated with UCSB), were some of the reasons why UCSB was on
the Arpanet so early.

There were no classes (that I'm aware of) that studied networking,
nor classes that used networking as a tool. In fact, the only
campus-wide network on campus was an Ungerman-Bass network used
to connect terminals to important nodes on campus. When I left UCSB
in 1985 the only WAN networking in place was DECNET between the
Physics Dept (where I was the computing manager) and SLAC.
Jim Frew, who sometimes posts on the list, could correct me if I'm wrong
about any of this. It was a long time ago.

Jon Forrest

[TUHS] Re: Berkeley CSRG Building

[Noel Chiappa]

    > From: Larry McVoy

{Moving this to COFF, as it's not UNIX-related. I'll have another reply there
as well, about the social media point.}

    > The amazing thing, to me, is I was a CS student in very early 1980's
    > and I had no idea of the history behind the arpanet.

I don't think that was that uncommon; at MIT (slightly earlier, I think -
-'74-'77 for me) the undergrad's weren't learning anything about networking
there either, then.

I think the reason is that there wasn't much to teach - in part because we
did not then know much about networking, and in part because it was not yet
crystal clear how important it would become (more below on that).

There was research going on in the area, but even at MIT one doesn't teach
(or didn't then; I don't know about now) on-going research subjects to
undergrads. MIT _did_ have, even then, a formal UROP ('undergrad research
opportunities') program, which allowed undergrads to be part of research
groups - a sheer genius idea - which in some fast-moving fields, like CS, was
an inestimable benefit to more forward undergrads in those fields.


I joined the CSR group at LCS in '77 because I had some operating system
ideas I wanted to work on; I had no idea at that point that they were doing
anything with networks. They took me on as the result of the sheerest chance;
they had just gotten some money from DARPA to build a LAN, and the interface
was going to be built for a UNIBUS PDP-11, and they needed diagnostics, etc
written; and they were all Multicians. I, by chance, knew PDP-11 assembler -
which none of them did - the MIT CS introductory course at that point taught
it. So the deal was that I'd help them with that, and I could use the machine
to explore my OS ideas in return.

Which never really happened; it fairly became clear to me that data
networking was going to have an enormous impact on the world, and at that
point it was also technically interesting, so I quickly got sucked into that
stuff. (I actually have a written document hiding in a file drawer somewhere
from 1978 or so, which makes it plain that that I'm not suffering 20-20
hindsight here, in talking about foreseeing the impact; I should dig it up.)

The future impact actually wasn't hard to foresee: looking at what printed
books had done to the world, and then telgraphs/telephones, and what
computers had already started to do at that point, it was clear that
combining them all was going to have an incredible impact (and we're still
adapting to it).


Learning about networking at the time was tricky. The ARPANET - well, NCP and
below - was pretty well documented in a couple of AFIPS papers (linked to at
the bottom here:

  https://gunkies.org/wiki/ARPANET

which I have a very vague memory I photocopied at the time out of the bound
AFIPS proceedings in the LCS library). The applications were only documented
in the RFC's.

(Speaking of which, at that level, the difference between the ARPANET and the
Internet was not very significant - it was only the internals, invisible to
the people who did 'application' protocols, that were completely different.
HTTP would probably run just fine on top of NCP, for instance.)

Anything past that, the start of the internet work, that, I picked up by i)
direct osmosis from other people in CSR who were starting to think about
networks - principally Dave Clark and Dave Reed - and then ii) from documents
prepared as part of the TCP/IP effort, which were distributed electronically.

Which is an interesting point; the ARPANET was a key tool in the internet
work. The most important aspect was email; non-stop discussion between the
widely separated groups who were part of the project. It also made document
distribution really easy (which had also been true of the latter stages of
the ARPANET project, with the RFC's). And of course it was also a long-haul
network that we used to tie the small internets at all the various sites
(BBN, SRI, ISI - and eventually MIT) into the larger Internet.

I hate to think about trying to do all that work on internets, and the
Internet, without the ARPANE there, as a tool.

	Noel

[TUHS] Re: Berkeley CSRG Building

[Al Kossow]

On 8/12/24 7:23 AM, Larry McVoy wrote:
> I was on UUCP, lm@cs.wisc.edu
> was not yet a thing for me.  All this history going on around me and I was
> clueless.

and I was at UW-Milw who had a connection through INHP-4
Weirdly, I ran into Rusty at a rest stop on I-80 in Jan 1984 heading west.

[TUHS] Re: Berkeley CSRG Building

[Larry McVoy]

On Mon, Aug 12, 2024 at 07:51:31AM -0400, Noel Chiappa wrote:
>     > From: "Erik E. Fair"
> 
>     > before that, ARPANET was connected
> 
> NOTE: "ARPANET" is _always_ (**ALWAYS**) used with the definite article
> ("the"). Don't take my word for it; check out, e.g.
> 
>   A History of the ARPANET: The First Decade
>   https://walden-family.com/bbn/arpanet-completion-report.pdf
>   https://apps.dtic.mil/sti/pdfs/ADA115440.pdf
> 
> (There are a lot more contemporaneous documents - AFIPS conference papers,
> etc - if anyone is interested in them.)
> 
> I was somewhat polite about it _this_ time.
> 
> 	Noel

The amazing thing, to me, is I was a CS student in very early 1980's and I
had no idea of the history behind the arpanet.  I was on UUCP, lm@cs.wisc.edu
was not yet a thing for me.  All this history going on around me and I was
clueless.

The executive summary in that first link is spot on,  They knew what they
had.  

Sadly, they didn't anticipate facebook and bots et al, but I'm not sure,
even now, that they could have done anything about that, a packet is a
packet whether it is for good or evil.
-- 
---
Larry McVoy           Retired to fishing          http://www.mcvoy.com/lm/boat

[TUHS] Re: Berkeley CSRG Building

[Will Senn]

[-- Attachment #1: Type: text/plain, Size: 812 bytes --]

On 8/12/24 06:51, Noel Chiappa wrote:
>      > From: "Erik E. Fair"
>
>      > before that, ARPANET was connected
>
> NOTE: "ARPANET" is _always_ (**ALWAYS**) used with the definite article
> ("the"). Don't take my word for it; check out, e.g.
>
>    A History of the ARPANET: The First Decade
>    https://walden-family.com/bbn/arpanet-completion-report.pdf
>    https://apps.dtic.mil/sti/pdfs/ADA115440.pdf
>
> (There are a lot more contemporaneous documents - AFIPS conference papers,
> etc - if anyone is interested in them.)
>
> I was somewhat polite about it _this_ time.
>
> 	Noel
Nice. the ARPANET... why do I think history and language fluidity will 
win the war of words? google code-shifting. Just typing that last 
sentence is enough to give grammarians everywhere tremors.

Great links though.

Will

[-- Attachment #2: Type: text/html, Size: 1445 bytes --]

[TUHS] Re: Berkeley CSRG Building

[Noel Chiappa]

    > From: "Erik E. Fair"

    > before that, ARPANET was connected

NOTE: "ARPANET" is _always_ (**ALWAYS**) used with the definite article
("the"). Don't take my word for it; check out, e.g.

  A History of the ARPANET: The First Decade
  https://walden-family.com/bbn/arpanet-completion-report.pdf
  https://apps.dtic.mil/sti/pdfs/ADA115440.pdf

(There are a lot more contemporaneous documents - AFIPS conference papers,
etc - if anyone is interested in them.)

I was somewhat polite about it _this_ time.

	Noel