|
Note 9
FROM: Ben Barker
I put in node 1 at UCLA in September. It came up and ran fine. I ran
my diagnostic for quite a while to make sure everything was intact, then
loaded the operational program. It came up and ran fine, but had
nothing to talk to. Mike Wingfield, at the time a UCLA grad student,
had designed an interface to the UCLA Sigma 7 in accordance with the
1822 spec that the BBN group had written. We plugged in his interface
and he loaded his diagnostic on the Sigma 7. It came up and ran
properly, but crashed after an hour or so. I suggested that it sounded
like a synchronizer problem, and we pored over his drawings until we
found the problem. (Mike had never learned about the synchronizer
problem, and believed that a flip flop would either be 1 or 0.) We
patched the problem and brought the interface back up. Worked fine from
there on. We waited a while to make sure everything kept running and
then went out for supper. It was still running fine the next morning.
I called Frank Heart to say we were done and I was coming home. He told
me to stay out there for a few days to make sure nothing happened.
Truett Thach and I spent a few days enjoying LA's pizza and pool halls,
then I came home.
A month later, Marty Thrope installed IMP 2 at SRI, and I went back to
UCLA to coordinate. As Marty brought up #2, I saw the console light go
out indicating that the line was up. I crosspatched from the UCLA TTY
to the SRI TTY and typed what was, in some sense, the first
communication across the net. I believe it was,
"MISAN-
'LO
/XOLOX"
25 years later, folks starting asking me what was the first message. I
wasn't sure, and called Marty to ask if he remembered. He said no. I
suggested that we could agree on something and as long as we stuck to
our stories, there was nobody who could question us, as there was nobody
else present at the time. He said no, the important thing that folks
should remember is that at the time it was all viewed as of so little
importance that nobody bothered making up something noteworthy or
recording what was sent.
It was a couple of weeks later, as I recall, that SRI got their host
connected to the IMP and Charlie Kline tried to log in. This was the
first host-to-host communication between distant nodes. Kleinrock and
his group, notably Kline, report this as the first communication across
the net. It was not.
The first communications between IMPs happened in Cambridge between node
0, the prototype, and node 1. But that was all in one room, so maybe it
didn't count as communication over the net. Similarly, the first
communication between people over IMPs happened in the same room and the
same configuration. The first communication between remote IMPs was
when Marty brought up IMP 2 at SRI and the two IMPs intercommunicated
their routing and status information. The first communication between
people over remotely-connected IMPs was the TTY-to-TTY conversation
between Marty and myself. The first host-to-host communication between
remotely connected IMPs was the login attempt so often recounted by
Kline.
I do not believe the login attempt was done over NCP. I am fairly
confident that NCP was not yet implemented, perhaps not yet specified.
I believe it was done over a kludge, like the software I wrote on the
Harvard PDP-10 when I built the 1822 interface for them. That kludge
allowed me to crosspatch to the PDP-10 from an IMP TTY and log in. The
software just took characters arriving from the IMP and shoved them down
a PTY, and vice-versa.
Ah, memory lane...
/Ben
I put in node 1 at UCLA in September. It came up and ran fine. I ran
my diagnostic for quite a while to make sure everything was intact, then
loaded the operational program. It came up and ran fine, but had
nothing to talk to. Mike Wingfield, at the time a UCLA grad student,
had designed an interface to the UCLA Sigma 7 in accordance with the
1822 spec that the BBN group had written. We plugged in his interface
and he loaded his diagnostic on the Sigma 7. It came up and ran
properly, but crashed after an hour or so. I suggested that it sounded
like a synchronizer problem, and we pored over his drawings until we
found the problem. (Mike had never learned about the synchronizer
problem, and believed that a flip flop would either be 1 or 0.) We
patched the problem and brought the interface back up. Worked fine from
there on. We waited a while to make sure everything kept running and
then went out for supper. It was still running fine the next morning.
I called Frank Heart to say we were done and I was coming home. He told
me to stay out there for a few days to make sure nothing happened.
Truett Thach and I spent a few days enjoying LA's pizza and pool halls,
then I came home.
A month later, Marty Thrope installed IMP 2 at SRI, and I went back to
UCLA to coordinate. As Marty brought up #2, I saw the console light go
out indicating that the line was up. I crosspatched from the UCLA TTY
to the SRI TTY and typed what was, in some sense, the first
communication across the net. I believe it was,
"MISAN-
'LO
/XOLOX"
25 years later, folks starting asking me what was the first message. I
wasn't sure, and called Marty to ask if he remembered. He said no. I
suggested that we could agree on something and as long as we stuck to
our stories, there was nobody who could question us, as there was nobody
else present at the time. He said no, the important thing that folks
should remember is that at the time it was all viewed as of so little
importance that nobody bothered making up something noteworthy or
recording what was sent.
It was a couple of weeks later, as I recall, that SRI got their host
connected to the IMP and Charlie Kline tried to log in. This was the
first host-to-host communication between distant nodes. Kleinrock and
his group, notably Kline, report this as the first communication across
the net. It was not.
The first communications between IMPs happened in Cambridge between node
0, the prototype, and node 1. But that was all in one room, so maybe it
didn't count as communication over the net. Similarly, the first
communication between people over IMPs happened in the same room and the
same configuration. The first communication between remote IMPs was
when Marty brought up IMP 2 at SRI and the two IMPs intercommunicated
their routing and status information. The first communication between
people over remotely-connected IMPs was the TTY-to-TTY conversation
between Marty and myself. The first host-to-host communication between
remotely connected IMPs was the login attempt so often recounted by
Kline.
I do not believe the login attempt was done over NCP. I am fairly
confident that NCP was not yet implemented, perhaps not yet specified.
I believe it was done over a kludge, like the software I wrote on the
Harvard PDP-10 when I built the 1822 interface for them. That kludge
allowed me to crosspatch to the PDP-10 from an IMP TTY and log in. The
software just took characters arriving from the IMP and shoved them down
a PTY, and vice-versa.
Ah, memory lane...
/Ben
Copyright 2012-2022 @ xbbn.org under a Creative Commons license.
