24 Oct
2021
24 Oct
'21
4:16 p.m.
Hello,
I'm working on setting up an emulated ARPANET using the original IMP
software recovered some years ago. It turns out, the greatest challenge
is finding the NCP software on the host side that implements the ARPANET
protocols. I have searched the TUHS archive and elsewhere, but all I
find for Unix is a copy of the PDP-11 Unix V6 NCP from Illinois.
Has any other NCP implementation for Unix survived? From old host
tables I think there may have been some VAXen online before the switch
to TCP/IP.
Best regards,
Lars Brinkhoff
24 Oct
24 Oct
5:24 p.m.
On Sun, Oct 24, 2021 at 12:57 PM Lars Brinkhoff <lars(a)nocrew.org> wrote:
Has any other NCP implementation for Unix survived?
From old host
tables I think there may have been some VAXen online before the switch
to TCP/IP.
Hmmm... I find that interesting because Stan Smith and I wrote the
original IP/TCP for the VAX in 1979/1980 when we were at Tektronix(in
BLISS-32, BTW -- which I gave back to CMU, and CMU added mailer and
eventual DNS support, and then distributed it as the Tek-CMU TCP/IP - DEC
eventually used it to start their in-house effort - I have that code on
9-track and may have it on a CD somewhere). When Stan and I started that
work, I had looked everywhere I could think to find an NCP implementation,
and never found one. Our Vax was connected to the Teklabs 11/70 running
V7 and the 3Com UNET stack that Bruce Bordan and crew wrote.
The reason was that IMP's ports were a rare commodity and free ones almost
non-existent. So when Vaxen started to roll out, the IMPs were
usually filed with PDP-10s at the research sites. Vaxen also tended to
have Xerox 3M Ethernet boards. I also don't know of anyone connecting a
DR-11B to an 1822 for a Vax - could have been done, but I don't know of
one. At both CMU and UCB, we used Xerox boards until the real ones from
3Com, Interlan and DEC showed up and that was not until 1981-82 and the IP
transition has already begun.
You tell me, but I was under the impression, that you folks at MIT did a
ChaosNet interface, IIRC, so there may have been some sort of conversion on
your LAN, but I really doubt there was a real NCP running.
FYI: CMU had Vax serial #1, and it was never on the ARPAnet. It was in the
same machine room as the IBM 360s, PDP-20s, and Univac 1108, not the CS
PDP-10s where the IMP was.
My guess is that people had Vaxen assigned in the host tables, but never
connected for lack of an NCP implementation. And until the Tek stack for
VMS and the BBN stack for 4.1 BSD, I don't think there were any Vaxen yet
on either the NCP based ARPAnet much less the later IP-based one.
Clem
ᐧ
ᐧ
6:32 p.m.
There were definitely as many PDP-11's (most running UNIX) as there were
DEC 10s in the glory days of the Arpanet. VAXes only rolled out toward
the end of the NCP era.
However, the last NCP host table shows this statistic for DEC machines
on the NCP Arpanet
VAX (UNIX): 58
VAX (VMS): 19
PDP11 (UNIX): 59
PDP11 (RSX): 6
PDP11 (MOS): 11
PDP11 (MINITS); 10
PDP10 (TOPS-20): 40
PDP10 (TOPS-10): 7
PDP10 (TENEX): 22
PDP10 (ITS): 4
PDP10 (WAITS): 3
I had all but forgotten about Local Host / Distant Host / Very Distant
Host 1822 protocols. I remember that BRL had a PDP-11/40 running ANTS
(ArpaNet Terminal Server out of University of Illinois). It got
replaced by an 11/34 running UNIX when the Arpanet went to long leaders
New Years 1981.
7:28 p.m.
Ron. Where did the back ncp come from?as I said I never saw it and we tried
to find one. Same for an IP implementation. That's why we wrote one. We
were 3coms first customer and I somewhere have the mailing evenlope marked
the 32 of December because they had a VC payment dependent on delivering
before end of year.
As for 11s. Yes that is true. A lot of them front ended large systems as
Bob pointed out but many were self supporting as you note.
On Sun, Oct 24, 2021 at 2:41 PM Ron Natalie <ron(a)ronnatalie.com> wrote:
There were definitely as many PDP-11's (most
running UNIX) as there were
DEC 10s in the glory days of the Arpanet. VAXes only rolled out toward
the end of the NCP era.
However, the last NCP host table shows this statistic for DEC machines on
the NCP Arpanet
VAX (UNIX): 58
VAX (VMS): 19
PDP11 (UNIX): 59
PDP11 (RSX): 6
PDP11 (MOS): 11
PDP11 (MINITS); 10
PDP10 (TOPS-20): 40
PDP10 (TOPS-10): 7
PDP10 (TENEX): 22
PDP10 (ITS): 4
PDP10 (WAITS): 3
I had all but forgotten about Local Host / Distant Host / Very Distant
Host 1822 protocols. I remember that BRL had a PDP-11/40 running ANTS
(ArpaNet Terminal Server out of University of Illinois). It got replaced
by an 11/34 running UNIX when the Arpanet went to long leaders New Years
1981.
--
Sent from a handheld expect more typos than usual
7:28 p.m.
On Sun, Oct 24, 2021 at 12:41 PM Ron Natalie <ron(a)ronnatalie.com> wrote:
There were definitely as many PDP-11's (most
running UNIX) as there were
DEC 10s in the glory days of the Arpanet. VAXes only rolled out toward
the end of the NCP era.
However, the last NCP host table shows this statistic for DEC machines on
the NCP Arpanet
VAX (UNIX): 58
PDP11(UNIX): 59
So as of 1980-12-31, you had 7th Edition and 32V in the wild. 3BSD was late
1979 or early 1980. 4BSD was late 1980.
The Columbus Unix was also running on VAXen, but I don't have great dates
for that and most (all?) of those machines
were confined to internal AT&T.
I recall reading, but don't recall where I read it, that the Urbana
networking was ported to 7th Edition. I tried to find
such source, but couldn't 18 months ago. 32V is basically the same as 7th
Edition, so would make a good candidate
for such a port, though 3BSD isn't so radically different, except in buffer
mapping, that it couldn't run there w/o much
effort. The VAX never had a V6 port that I've read about anywhere, though
I'll defer to others with better sources.
I'd bet some $$$ that BBN did the port, but I have no good source for that
either.
Warner
8:25 p.m.
Sorry for typo and top posting on my iPhone for a few days as I'm traveling
W/O a computer.
Some thoughts.
There a number of pdp11 stacks for ip and I assume we the ncp but they were
not all sourced by bbn. A number were in assembler
As Noel pointed out the first widely distributed Unix implementation for
the ncp was the uici version. But there was a Rand stack around the same
time and III ty hink Holmgren ended up at UCSB after his time at UICI. Im
fairly sure there was cross polinartion but I don't know how much.
In the IP space the 11 assembler version was the first u knew about. The
first C version was the Haverty stack that became what knew of as the
portable IP stack originally for the Nova. Rob Gurwitz (working) for Jack
did the 4.1 code and was the original official Vax stack for DARPA. Joy's
reimplementation for 4.1a was sort of unsanctioned and caused a bit of a
rift as it was not a complete rewrite and officially BBN was doing the
network support for the VAX for DARPA. CSRG was just supposed to being
doing base OS.
But the joy version is derrived from the Gurwitz/Havert work - For instance
the famous mbuf code came from the DG Nova implementation a few years
earlier. [I was once told mbufs came from Brown, but Jack said on the IH
mailing list in the last year or so, he worked on them before Rob started
his work].
Stan and I only had assembler and BLISS and Fortran on the Vax under VMS so
I started at the DG version when we did something similar in the Tek
version [it was not as elegant or as complicated and probably not as good].
As for other NCPs, PARC had MAXC on the net, but I thought it had
originally a DG Nova front end that was replaced with an Alto. But the NCP
ran as Nova code in BCPL in both.
On Sun, Oct 24, 2021 at 3:30 PM Warner Losh <imp(a)bsdimp.com> wrote:
On Sun, Oct 24, 2021 at 12:41 PM Ron Natalie <ron(a)ronnatalie.com> wrote:
Sent from a handheld expect more typos than usual
There were definitely as many PDP-11's (most
running UNIX) as there were
DEC 10s in the glory days of the Arpanet. VAXes only rolled out toward
the end of the NCP era.
However, the last NCP host table shows this statistic for DEC machines on
the NCP Arpanet
VAX (UNIX): 58
PDP11(UNIX): 59
So as of 1980-12-31, you had 7th Edition and 32V in the wild. 3BSD was
late 1979 or early 1980. 4BSD was late 1980.
The Columbus Unix was also running on VAXen, but I don't have great dates
for that and most (all?) of those machines
were confined to internal AT&T.
I recall reading, but don't recall where I read it, that the Urbana
networking was ported to 7th Edition. I tried to find
such source, but couldn't 18 months ago. 32V is basically the same as 7th
Edition, so would make a good candidate
for such a port, though 3BSD isn't so radically different, except in
buffer mapping, that it couldn't run there w/o much
effort. The VAX never had a V6 port that I've read about anywhere, though
I'll defer to others with better sources.
I'd bet some $$$ that BBN did the port, but I have no good source for that
either.
Warner
--
25 Oct
25 Oct
9:26 a.m.
Clem Cole wrote:
That seems consistent with host tables I see. According to those, there
were no VAXen on the ARPANET in 1980. However, there were a a dozen or
so in mid 1981; mostly VMS. (I'd also like to ask about NCP for VMS,
but this is TUHS.)
In early 1982 the number of VAXen has doubled, and out of those 10 are
running Unix including prominent sites like BBN, CCA, PURDUE, RAND, UCB.
Here's the rub. Some hosts may have jumped the TCP/IP gun ahead of the
1/1 1983 flag day. The host tables don't say. Could it be that all
those VAXen were running experimental TCP/IP in January 1982?
Has any other
NCP implementation for Unix survived? From old host
tables I think there may have been some VAXen online before the
switch to TCP/IP.
Hmmm... I find that interesting because Stan Smith and I wrote the
original IP/TCP for the VAX in 1979/1980 when we were at Tektronix
[...]. When Stan and I started that work, I had looked everywhere I
could think to find an NCP implementation, and never found one. You tell me, but I was under the impression, that you
folks at MIT did
a ChaosNet interface, IIRC, so there may have been some sort of
conversion on your LAN, but I really doubt there was a real NCP
running.
The MIT host tables do show the Chaos networks, but I don't include
those in the count.
913
days inactive
914
days old
6 comments
4 participants
participants (4)
-
Clem Cole -
Lars Brinkhoff -
Ron Natalie -
Warner Losh