8 Mar
2023
8 Mar
'23
9:47 p.m.
I have been planning for a long time to post a small reflection when I
reached 1000 nodes registered in the hecnet nodename database.
This happened tonight. I think it is a pretty cool thing. There are now
1000 nodenames registered in this small hobby DECnet. I guess you could
say it's actually not that small.
However, I know that rather few machines are actually online, and it
might even be that the majority of nodes registered have never been
online. My guess is that maybe 5% of the registered machines are usually
online. But I do not have any hard data to back this up.
HECnet started out about 20 years ago from a desire I had to hook up a
PDP-11 I had at home, to some machines at my university, using DECnet.
At the time, I didn't have any TCP/IP for RSX, and the only way to get
any kind of networking was to try and come up with some way of getting
DECnet up.
My first implementation basically just forwarded a serial port
communication between two Unix machines. And on each end I then hooked
that serial port into a PDP-11 running RSX, and used DDCMP for the
actual DECnet link.
This worked, but was obviously not that fast, as the serial ports were
limited to 9600 bps.
After a year or so, I figured I could instead write a small program that
would forward ethernet packets. Using UDP I basically had the same
property as a local ethernet, but it could be located somewhere pretty
far away. From a DECnet point of view, it would appear as if they were
on the same ethernet segment, while in reality they were nowhere near.
As DECnet have pretty long timeouts on things, it turned out this worked
without a hitch, and I could achieve much better throughput.
Not long after that, the first other users were hooked up to HECnet as
well. This was maybe around 2003 or so.
Another data point is that there are 113 different persons that have one
or more nodenames registered.
Happy milestone, everyone, and thanks for being around.
Johnny
--
Johnny Billquist || "I'm on a bus
|| on a psychedelic trip
email: bqt(a)softjar.se || Reading murder books
pdp is alive! || tryin' to stay hip" - B. Idol
8 Mar
8 Mar
10:12 p.m.
Very cool. Congratulations to the milestone
On Wed, Mar 8, 2023 at 12:47 Johnny Billquist <bqt(a)softjar.se> wrote:
I have been planning for a long time to post a small
reflection when I
reached 1000 nodes registered in the hecnet nodename database.
This happened tonight. I think it is a pretty cool thing. There are now
1000 nodenames registered in this small hobby DECnet. I guess you could
say it's actually not that small.
However, I know that rather few machines are actually online, and it
might even be that the majority of nodes registered have never been
online. My guess is that maybe 5% of the registered machines are usually
online. But I do not have any hard data to back this up.
HECnet started out about 20 years ago from a desire I had to hook up a
PDP-11 I had at home, to some machines at my university, using DECnet.
At the time, I didn't have any TCP/IP for RSX, and the only way to get
any kind of networking was to try and come up with some way of getting
DECnet up.
My first implementation basically just forwarded a serial port
communication between two Unix machines. And on each end I then hooked
that serial port into a PDP-11 running RSX, and used DDCMP for the
actual DECnet link.
This worked, but was obviously not that fast, as the serial ports were
limited to 9600 bps.
After a year or so, I figured I could instead write a small program that
would forward ethernet packets. Using UDP I basically had the same
property as a local ethernet, but it could be located somewhere pretty
far away. From a DECnet point of view, it would appear as if they were
on the same ethernet segment, while in reality they were nowhere near.
As DECnet have pretty long timeouts on things, it turned out this worked
without a hitch, and I could achieve much better throughput.
Not long after that, the first other users were hooked up to HECnet as
well. This was maybe around 2003 or so.
Another data point is that there are 113 different persons that have one
or more nodenames registered.
Happy milestone, everyone, and thanks for being around.
Johnny
--
Johnny Billquist || "I'm on a bus
|| on a psychedelic trip
email: bqt(a)softjar.se || Reading murder books
pdp is alive! || tryin' to stay hip" - B. Idol
_______________________________________________
HECnet mailing list -- hecnet(a)lists.dfupdate.se
To unsubscribe send an email to hecnet-leave(a)lists.dfupdate.se
--
Sent from Gmail because I have to :(
10:24 p.m.
I remember that serial program. I ran it on a DEC MIPS box running NetBSD and had the
serial port connected to a VAX 5000/300. Both sitting in my livingroom.
It was the coolest thing ever. We've come a very long way since then!
I can't tell you how much being a part of this has meant to me over the years.
It's always been a fun thing to do but the best part is the people I've met along
the way.
Thank you for starting this.
On Mar 8, 2023, 20:47 +0000, Johnny Billquist <bqt(a)softjar.se>se>, wrote:
I have been planning for a long time to post a small
reflection when I
reached 1000 nodes registered in the hecnet nodename database.
This happened tonight. I think it is a pretty cool thing. There are now
1000 nodenames registered in this small hobby DECnet. I guess you could
say it's actually not that small.
However, I know that rather few machines are actually online, and it
might even be that the majority of nodes registered have never been
online. My guess is that maybe 5% of the registered machines are usually
online. But I do not have any hard data to back this up.
HECnet started out about 20 years ago from a desire I had to hook up a
PDP-11 I had at home, to some machines at my university, using DECnet.
At the time, I didn't have any TCP/IP for RSX, and the only way to get
any kind of networking was to try and come up with some way of getting
DECnet up.
My first implementation basically just forwarded a serial port
communication between two Unix machines. And on each end I then hooked
that serial port into a PDP-11 running RSX, and used DDCMP for the
actual DECnet link.
This worked, but was obviously not that fast, as the serial ports were
limited to 9600 bps.
After a year or so, I figured I could instead write a small program that
would forward ethernet packets. Using UDP I basically had the same
property as a local ethernet, but it could be located somewhere pretty
far away. From a DECnet point of view, it would appear as if they were
on the same ethernet segment, while in reality they were nowhere near.
As DECnet have pretty long timeouts on things, it turned out this worked
without a hitch, and I could achieve much better throughput.
Not long after that, the first other users were hooked up to HECnet as
well. This was maybe around 2003 or so.
Another data point is that there are 113 different persons that have one
or more nodenames registered.
Happy milestone, everyone, and thanks for being around.
Johnny
--
Johnny Billquist || "I'm on a bus
|| on a psychedelic trip
email: bqt(a)softjar.se || Reading murder books
pdp is alive! || tryin' to stay hip" - B. Idol
_______________________________________________
HECnet mailing list -- hecnet(a)lists.dfupdate.se
To unsubscribe send an email to hecnet-leave(a)lists.dfupdate.se
9 Mar
9 Mar
1:55 a.m.
You might be the only one apart from me who ever ran that serial line
solution, Brian. :-)
But I agree with what pretty much all have been saying. It's a lot of
fun, and it have enabled similar (weird) minded people to connect.
Thanks to all of you who are here. You are all an equal part in the fun
of this.
Johnny
On 2023-03-08 22:24, Brian Hechinger wrote:
I remember that serial program. I ran it on a DEC MIPS
box running
NetBSD and had the serial port connected to a VAX 5000/300. Both sitting
in my livingroom.
It was the coolest thing ever. We've come a very long way since then!
I can't tell you how much being a part of this has meant to me over the
years. It's always been a fun thing to do but the best part is the
people I've met along the way.
Thank you for starting this.
On Mar 8, 2023, 20:47 +0000, Johnny Billquist <bqt(a)softjar.se>se>, wrote:
I have been planning for a long time to post a small reflection when I
reached 1000 nodes registered in the hecnet nodename database.
This happened tonight. I think it is a pretty cool thing. There are now
1000 nodenames registered in this small hobby DECnet. I guess you could
say it's actually not that small.
However, I know that rather few machines are actually online, and it
might even be that the majority of nodes registered have never been
online. My guess is that maybe 5% of the registered machines are usually
online. But I do not have any hard data to back this up.
HECnet started out about 20 years ago from a desire I had to hook up a
PDP-11 I had at home, to some machines at my university, using DECnet.
At the time, I didn't have any TCP/IP for RSX, and the only way to get
any kind of networking was to try and come up with some way of getting
DECnet up.
My first implementation basically just forwarded a serial port
communication between two Unix machines. And on each end I then hooked
that serial port into a PDP-11 running RSX, and used DDCMP for the
actual DECnet link.
This worked, but was obviously not that fast, as the serial ports were
limited to 9600 bps.
After a year or so, I figured I could instead write a small program that
would forward ethernet packets. Using UDP I basically had the same
property as a local ethernet, but it could be located somewhere pretty
far away. From a DECnet point of view, it would appear as if they were
on the same ethernet segment, while in reality they were nowhere near.
As DECnet have pretty long timeouts on things, it turned out this worked
without a hitch, and I could achieve much better throughput.
Not long after that, the first other users were hooked up to HECnet as
well. This was maybe around 2003 or so.
Another data point is that there are 113 different persons that have one
or more nodenames registered.
Happy milestone, everyone, and thanks for being around.
Johnny
--
Johnny Billquist || "I'm on a bus
|| on a psychedelic trip
email: bqt(a)softjar.se || Reading murder books
pdp is alive! || tryin' to stay hip" - B. Idol
_______________________________________________
HECnet mailing list -- hecnet(a)lists.dfupdate.se
To unsubscribe send an email to hecnet-leave(a)lists.dfupdate.se
_______________________________________________
HECnet mailing list -- hecnet(a)lists.dfupdate.se
To unsubscribe send an email to hecnet-leave(a)lists.dfupdate.se
--
Johnny Billquist || "I'm on a bus
|| on a psychedelic trip
email: bqt(a)softjar.se || Reading murder books
pdp is alive! || tryin' to stay hip" - B. Idol
8 Mar
8 Mar
10:28 p.m.
Congratulations! Happy, kilo-node day!
On 08/03/2023 20:47, Johnny Billquist wrote:
I have been planning for a long time to post a small
reflection when I
reached 1000 nodes registered in the hecnet nodename database.
This happened tonight. I think it is a pretty cool thing. There are
now 1000 nodenames registered in this small hobby DECnet. I guess you
could say it's actually not that small.
However, I know that rather few machines are actually online, and it
might even be that the majority of nodes registered have never been
online. My guess is that maybe 5% of the registered machines are
usually online. But I do not have any hard data to back this up.
HECnet started out about 20 years ago from a desire I had to hook up a
PDP-11 I had at home, to some machines at my university, using DECnet.
At the time, I didn't have any TCP/IP for RSX, and the only way to get
any kind of networking was to try and come up with some way of getting
DECnet up.
My first implementation basically just forwarded a serial port
communication between two Unix machines. And on each end I then hooked
that serial port into a PDP-11 running RSX, and used DDCMP for the
actual DECnet link.
This worked, but was obviously not that fast, as the serial ports were
limited to 9600 bps.
After a year or so, I figured I could instead write a small program
that would forward ethernet packets. Using UDP I basically had the
same property as a local ethernet, but it could be located somewhere
pretty far away. From a DECnet point of view, it would appear as if
they were on the same ethernet segment, while in reality they were
nowhere near. As DECnet have pretty long timeouts on things, it turned
out this worked without a hitch, and I could achieve much better
throughput.
Not long after that, the first other users were hooked up to HECnet as
well. This was maybe around 2003 or so.
Another data point is that there are 113 different persons that have
one or more nodenames registered.
Happy milestone, everyone, and thanks for being around.
Johnny
10:28 p.m.
Well done Jonny.
While many of us probably don’t have the time to fire up the hardware for connecting, I
enjoy following the activity of those who do.
Regards, John
New Zealand
On 9/03/2023, at 09:47, Johnny Billquist
<bqt(a)softjar.se> wrote:
I have been planning for a long time to post a small reflection when I reached 1000
nodes registered in the hecnet nodename database.
This happened tonight. I think it is a pretty cool thing. There are now 1000 nodenames
registered in this small hobby DECnet. I guess you could say it's actually not that
small.
However, I know that rather few machines are actually online, and it might even be that
the majority of nodes registered have never been online. My guess is that maybe 5% of the
registered machines are usually online. But I do not have any hard data to back this up.
HECnet started out about 20 years ago from a desire I had to hook up a PDP-11 I had at
home, to some machines at my university, using DECnet. At the time, I didn't have any
TCP/IP for RSX, and the only way to get any kind of networking was to try and come up with
some way of getting DECnet up.
My first implementation basically just forwarded a serial port communication between two
Unix machines. And on each end I then hooked that serial port into a PDP-11 running RSX,
and used DDCMP for the actual DECnet link.
This worked, but was obviously not that fast, as the serial ports were limited to 9600
bps.
After a year or so, I figured I could instead write a small program that would forward
ethernet packets. Using UDP I basically had the same property as a local ethernet, but it
could be located somewhere pretty far away. From a DECnet point of view, it would appear
as if they were on the same ethernet segment, while in reality they were nowhere near. As
DECnet have pretty long timeouts on things, it turned out this worked without a hitch, and
I could achieve much better throughput.
Not long after that, the first other users were hooked up to HECnet as well. This was
maybe around 2003 or so.
Another data point is that there are 113 different persons that have one or more
nodenames registered.
Happy milestone, everyone, and thanks for being around.
Johnny
--
Johnny Billquist || "I'm on a bus
|| on a psychedelic trip
email: bqt(a)softjar.se || Reading murder books
pdp is alive! || tryin' to stay hip" - B. Idol
_______________________________________________
HECnet mailing list -- hecnet(a)lists.dfupdate.se
To unsubscribe send an email to hecnet-leave(a)lists.dfupdate.se
10:28 p.m.
Hi Johnny,
Congrats and thanks for giving us all something fun to play with, and
especially so over the past few years. My nodes are up and down on no
particular schedule, but I and I'm sure most others here greatly look
forward to the inevitable questions and surprises arriving from the HECnet
mailing list and its assortment of like minded community members.
You've done a very good thing... :)
Cheers,
Brian
On Wed, Mar 8, 2023 at 3:47 PM Johnny Billquist <bqt(a)softjar.se> wrote:
I have been planning for a long time to post a small
reflection when I
reached 1000 nodes registered in the hecnet nodename database.
This happened tonight. I think it is a pretty cool thing. There are now
1000 nodenames registered in this small hobby DECnet. I guess you could
say it's actually not that small.
However, I know that rather few machines are actually online, and it
might even be that the majority of nodes registered have never been
online. My guess is that maybe 5% of the registered machines are usually
online. But I do not have any hard data to back this up.
HECnet started out about 20 years ago from a desire I had to hook up a
PDP-11 I had at home, to some machines at my university, using DECnet.
At the time, I didn't have any TCP/IP for RSX, and the only way to get
any kind of networking was to try and come up with some way of getting
DECnet up.
My first implementation basically just forwarded a serial port
communication between two Unix machines. And on each end I then hooked
that serial port into a PDP-11 running RSX, and used DDCMP for the
actual DECnet link.
This worked, but was obviously not that fast, as the serial ports were
limited to 9600 bps.
After a year or so, I figured I could instead write a small program that
would forward ethernet packets. Using UDP I basically had the same
property as a local ethernet, but it could be located somewhere pretty
far away. From a DECnet point of view, it would appear as if they were
on the same ethernet segment, while in reality they were nowhere near.
As DECnet have pretty long timeouts on things, it turned out this worked
without a hitch, and I could achieve much better throughput.
Not long after that, the first other users were hooked up to HECnet as
well. This was maybe around 2003 or so.
Another data point is that there are 113 different persons that have one
or more nodenames registered.
Happy milestone, everyone, and thanks for being around.
Johnny
--
Johnny Billquist || "I'm on a bus
|| on a psychedelic trip
email: bqt(a)softjar.se || Reading murder books
pdp is alive! || tryin' to stay hip" - B. Idol
_______________________________________________
HECnet mailing list -- hecnet(a)lists.dfupdate.se
To unsubscribe send an email to hecnet-leave(a)lists.dfupdate.se
10:36 p.m.
Johnny,
Congratulations on creating and maintaining this fun network. I’ve immensely enjoyed being
part of it, even though I don’t get nearly as much time to play as I’d like.
Here’s to the next 1000 nodes!
Ian
On Mar 8, 2023, at 12:47 PM, Johnny Billquist
<bqt(a)softjar.se> wrote:
I have been planning for a long time to post a small reflection when I reached 1000 nodes
registered in the hecnet nodename database.
This happened tonight. I think it is a pretty cool thing. There are now 1000 nodenames
registered in this small hobby DECnet. I guess you could say it's actually not that
small.
However, I know that rather few machines are actually online, and it might even be that
the majority of nodes registered have never been online. My guess is that maybe 5% of the
registered machines are usually online. But I do not have any hard data to back this up.
HECnet started out about 20 years ago from a desire I had to hook up a PDP-11 I had at
home, to some machines at my university, using DECnet. At the time, I didn't have any
TCP/IP for RSX, and the only way to get any kind of networking was to try and come up with
some way of getting DECnet up.
My first implementation basically just forwarded a serial port communication between two
Unix machines. And on each end I then hooked that serial port into a PDP-11 running RSX,
and used DDCMP for the actual DECnet link.
This worked, but was obviously not that fast, as the serial ports were limited to 9600
bps.
After a year or so, I figured I could instead write a small program that would forward
ethernet packets. Using UDP I basically had the same property as a local ethernet, but it
could be located somewhere pretty far away. From a DECnet point of view, it would appear
as if they were on the same ethernet segment, while in reality they were nowhere near. As
DECnet have pretty long timeouts on things, it turned out this worked without a hitch, and
I could achieve much better throughput.
Not long after that, the first other users were hooked up to HECnet as well. This was
maybe around 2003 or so.
Another data point is that there are 113 different persons that have one or more
nodenames registered.
Happy milestone, everyone, and thanks for being around.
Johnny
--
Johnny Billquist || "I'm on a bus
|| on a psychedelic trip
email: bqt(a)softjar.se || Reading murder books
pdp is alive! || tryin' to stay hip" - B. Idol
_______________________________________________
HECnet mailing list -- hecnet(a)lists.dfupdate.se
To unsubscribe send an email to hecnet-leave(a)lists.dfupdate.se
10:39 p.m.
On Mar 8, 2023, at 3:47 PM, Johnny Billquist
<bqt(a)softjar.se> wrote:
I have been planning for a long time to post a small reflection when I reached 1000 nodes
registered in the hecnet nodename database.
This happened tonight. I think it is a pretty cool thing. There are now 1000 nodenames
registered in this small hobby DECnet. I guess you could say it's actually not that
small.
However, I know that rather few machines are actually online, and it might even be that
the majority of nodes registered have never been online. My guess is that maybe 5% of the
registered machines are usually online. But I do not have any hard data to back this up.
That's a great milestone, and you deserve large amounts of credit for making it a
reality.
I guess I need to add some statistics to the HECnet mapper, showing the number of nodes
seen as "up". And perhaps a graph plotting that number over time. Maybe with
options to span various amounts of time, the way stock price charts do...
paul
9 Mar
9 Mar
12:50 a.m.
On 3/8/23 16:39, Paul Koning wrote:
That's a great milestone, and you deserve large
amounts of credit for making it a reality.
Agreed 100%!
I guess I need to add some statistics to the HECnet
mapper, showing the number of nodes seen as "up". And perhaps a graph plotting
that number over time. Maybe with options to span various amounts of time, the way stock
price charts do...
That would be good. What sort of polling interval do you think you
could do with that?
I ask because all of LSSM's DECnet-capable machines (which are all
real machines) are on HECnet whenever they're running, but it's usually
only for a few hours at a time, either when we're working on stuff or
when there are visitors at the museum.
Which reminds me, I need to get those updated in the node database.
-Dave
--
Dave McGuire, AK4HZ
New Kensington, PA
1:42 a.m.
On Mar 8, 2023, at 6:50 PM, Dave McGuire
<mcguire(a)neurotica.com> wrote:
On 3/8/23 16:39, Paul Koning wrote:
The polling interval is 24 hours, but that doesn't tell the whole story. A number of
routers are configured to send topology change events to the mapper, and it reacts to
those by doing an incremental scan (with a 15 second holdoff). So in general, whenever
you turn a node or circuit on or off, the map should react right away. You can confirm
this, probably most easily by looking at the "map data" table (a link off the
map display).
If incremental update isn't working for your nodes, a possible reason would be that
there isn't an event source in the area. Since the topology changes that the mapper
reacts to are all level 1 events, they depend on a router in that same area to let me
know. So for changes in an area not covered by an event sender, the 24 hour poll cycle
applies. At the top of the map display this is shown: "Network data last
updated" is the full scan, and "Last incremental change" is the most recent
event driven scan of just the reported spot.
I have 11 event senders right now, in a couple of cases several per area (good for
redundancy) but clearly not all areas are covered. Those with routers that can send
DECnet events to a "monitor sink" interested to volunteer should drop me a note,
I can supply the desired setup. You can see the currently connected senders at the bottom
of akdesign.dyndns.org:8080/nsp/status
paul
That's a great milestone, and you deserve
large amounts of credit for making it a reality.
Agreed 100%!
I guess I need to add some statistics to the
HECnet mapper, showing the number of nodes seen as "up". And perhaps a graph
plotting that number over time. Maybe with options to span various amounts of time, the
way stock price charts do...
That would be good. What sort of polling interval do you think you could do with that?
I ask because all of LSSM's DECnet-capable machines (which are all real machines)
are on HECnet whenever they're running, but it's usually only for a few hours at a
time, either when we're working on stuff or when there are visitors at the museum.
8 Mar
8 Mar
10:42 p.m.
Wow! Ignoring how many nodes are usually up, I agree that HECnet is hardly a small DECnet
deployment. I bet it's pretty large compared to a lot of "real" DECnet
deployments from back in the day. Thanks for creating HECnet and continuing to grow it.
This news brought a nice smile to my face today.
--
Mark J. Blair <nf6x(a)nf6x.net>
Blog: https://www.nf6x.net
Git: https://gitlab.com/users/NF6X/groups
HECnet: DOGPAK::MBLAIR
12 Mar
12 Mar
2:32 a.m.
It's certainly far larger than CCnet ever was. That was CMU, Columbia,
Case Western, Stevens and a few others.
The thing that was gigantic was BITNET, back in the day.
On 3/8/23 4:42 PM, Mark J. Blair wrote:
Wow! Ignoring how many nodes are usually up, I agree
that HECnet is hardly a small DECnet deployment. I bet it's pretty large compared to a
lot of "real" DECnet deployments from back in the day. Thanks for creating
HECnet and continuing to grow it. This news brought a nice smile to my face today.
3:32 a.m.
Thomas DeBellis <tommytimesharing(a)gmail.com>
wrote:
The thing that was gigantic was BITNET, back in the day.
How big was BITNET? When I was at DEC Easynet had around 5000 nodes. I heard that it
topped out at somewhere around 50,000 in the early 90s, but that was after my time.
Bob
3:47 a.m.
Columbia was the BITNET/Internet/CCnet gateway for the first third of
the 1980's. I was the systems programmer who maintained the 20's IBM
RJE entry software (IBMSPL) as well as the DN60 (PDP-11) front end HASP
and bi-sync drivers, one of the reasons being that I knew 360 assembler
and PDP-11 macro.
At the time, I believe BITNET had some absurd number of IBM 'hosts'; I
recall the number being something like 12,000. Whatever it was, it was
eye popping. I think it was world-wide.
This was pre-SNA/LU 6.2, by the way, so they all looked like RJE
workstations and communicated by 'punching' decks of cards to each
other. So there were no interactive services, but a large amount of
work was done via batch (still is via TSO), so most of the time you
didn't notice any real difference.
Personally, I vastly preferred editing my JCL with Emacs instead of
Xedit. If it were a choice between TECO and Xedit, it depended on what
kind of mood I was in.
------------------------------------------------------------------------
On 3/11/23 9:32 PM, Robert Armstrong wrote:
How big was BITNET? When I was at DEC Easynet had around 5000 nodes. I heard that
it topped out at somewhere around 50,000 in the early 90s, but that was after my time.
>
> ------------------------------------------------------------------------
> Thomas DeBellis<tommytimesharing(a)gmail.com> wrote:
>
> The thing that was gigantic was BITNET, back in the day.
4:26 a.m.
On 3/11/2023 9:47 PM, Thomas DeBellis wrote:
Columbia was the BITNET/Internet/CCnet gateway for the
first third of
the 1980's. I was the systems programmer who maintained the 20's IBM
RJE entry software (IBMSPL) as well as the DN60 (PDP-11) front end HASP
and bi-sync drivers, one of the reasons being that I knew 360 assembler
and PDP-11 macro.
There is a global hobbyist BITNET II network, with the central host
called IRLEARN in Ireland. IMPVAX is connected to it as well as HECnet.
There are quite a few DEC machines on that network, too. It is fun to
send commands and files and do a bunch of Q SYS and CPQ XYZ.
$ send/command irlearn "q sys"
$
(IRLEARN) - LINK ADMIN INACTIVE
$
(IRLEARN) - LINK IPVAX INACTIVE
$
(IRLEARN) - LINK IEGBOX INACTIVE
$
(IRLEARN) - LINK CCVAX INACTIVE
$
(IRLEARN) - LINK IRTCCARL INACTIVE
$
(IRLEARN) - LINK IRUCCIBM INACTIVE
$
(IRLEARN) - LINK UKACRL INACTIVE
$
(IRLEARN) - LINK PRTRSCS INACTIVE
$
(IRLEARN) - LINK BTP CONNECT -- LINE 038 NOH
(IRLEARN) - LINK CEIRE INACTIVE
(IRLEARN) - LINK SCOTTY INACTIVE
$
(IRLEARN) - LINK KEWTORIA CONNECT -- LINE 040 NOH
(IRLEARN) - LINK KEWRONI CONNECT -- LINE 041 NOH
(IRLEARN) - LINK AEARN CONNECT -- LINE 042 NOH
(IRLEARN) - LINK MOSHIX ACTIVE -- LINE 043 NOH
(IRLEARN) - LINK IMPVAX CONNECT -- LINE 044 NOH
(IRLEARN) - LINK NCCMAX CONNECT -- LINE 045 NOH
(IRLEARN) - LINK N00DLE CONNECT -- LINE 046 NOH
(IRLEARN) - LINK ETEXAS INACTIVE
(IRLEARN) - LINK VMF CONNECT -- LINE 048 NOH
(IRLEARN) - LINK FRIWVM01 CONNECT -- LINE 049 NOH
(IRLEARN) - LINK WABMVS CONNECT -- LINE 04A NOH
(IRLEARN) - LINK PJJV0 CONNECT -- LINE 04B NOH
(IRLEARN) - LINK MISERVM INACTIVE
(IRLEARN) - LINK THCHVM01 INACTIVE
(IRLEARN) - LINK CEARN CONNECT -- LINE 04E NOH
(IRLEARN) - LINK TANJPVM1 INACTIVE
(IRLEARN) - LINK CRNI01 ACTIVE -- LINE 04F NOH
(IRLEARN) - LINK FAKE ACTIVE -- LINE 036 NOH
At the time, I believe BITNET had some absurd number of IBM 'hosts'; I
recall the number being something like 12,000. Whatever it was, it was
eye popping. I think it was world-wide.
This was pre-SNA/LU 6.2, by the way, so they all looked like RJE
workstations and communicated by 'punching' decks of cards to each
other. So there were no interactive services, but a large amount of
work was done via batch (still is via TSO), so most of the time you
didn't notice any real difference.
Personally, I vastly preferred editing my JCL with Emacs instead of
Xedit. If it were a choice between TECO and Xedit, it depended on what
kind of mood I was in.
------------------------------------------------------------------------
On 3/11/23 9:32 PM, Robert Armstrong wrote:
How big was BITNET? When I was at DEC Easynet had around 5000 nodes. I heard that
it topped out at somewhere around 50,000 in the early 90s, but that was after my time.
>
> ------------------------------------------------------------------------
> Thomas DeBellis<tommytimesharing(a)gmail.com> wrote:
>
> The thing that was gigantic was BITNET, back in the day.
_______________________________________________
HECnet mailing list -- hecnet(a)lists.dfupdate.se
To unsubscribe send an email to hecnet-leave(a)lists.dfupdate.se
4:26 a.m.
On 2023-03-12 03:32, Robert Armstrong wrote:
the early 90s, but that was after my
time.
When were you at DEC? I was there in 1986, and Easynet was definitely
more than 5000 nodes by then.
According to Wikipedia, Bitnet was around 3000 nodes at it's peak in
1991. I'd say Easynet was easily way larger than Bitnet.
According to Wikipedia (again), Easynet was about 54000 nodes in 1990,
and about 15000 in 1987.
Johnny
--
Johnny Billquist || "I'm on a bus
|| on a psychedelic trip
email: bqt(a)softjar.se || Reading murder books
pdp is alive! || tryin' to stay hip" - B. Idol
Thomas
DeBellis <tommytimesharing(a)gmail.com> wrote:
The thing that was gigantic was BITNET, back in the day.
How big was BITNET? When I was at DEC Easynet had around 5000 nodes. I heard that
it topped out at somewhere around 50,000 in
9:05 p.m.
Gee, I thought he was asking me...
For what it's worth, I was with DEC in the Galaxy group from late 1978
to late 1980. I worked on the File Finder project, which was an online
database to keep track of backup tapes and the contents therein.
So if you zorched a file and wanted to restore it, you'd run File Finder
and find out what tape (or tapes) it was on and away you went. An
earlier version had been written in COBOL, which was using fixed length
records that were scanned sequentially. It rapidly blew up disk space
as well as having order O(N) CPU usage.
We got as far as designing a b-tree structured database, partly because
of the discovery process involved but mostly because we were given some
very poor implementation restrictions. It's too bad; it could have
easily been integrated into Galaxy, so once you found what you wanted,
you could get the mount request in and trigger the archive system's
restore process, which would have been very seamless. These days, that
kind of functionality is found everywhere (Veritas, Time Machine, Etc.)
I can't remember my badge number, but I still have it. Somewhere... I
can' barely remember what building I was in Marlboro. I think #3? (the
one on the right hand side closest to manufacturing and the upper
parking lot). I do remember the Galaxy group's machine, 2116 ("The Big
Orange Welcomes You") and, oddly enough, my sign-on id, "DEBELLIS" :-)
------------------------------------------------------------------------
On 3/12/23 10:56 AM, Robert Armstrong wrote:
I quit in 1985. Badge number 147367 (still have it memorized :)
Bob
> ------------------------------------------------------------------------
> Johnny Billquist<bqt(a)softjar.se> wrote:
> When were you at DEC?
9:11 p.m.
I was actually asking Bob, but I didn't know you were also at DEC. Cool.
What you describe sounds very similar to what offline files in TOPS-20
ended up like.
Anyway, I was at DEC in 1986, in Stockholm. Working in IS, mostly taking
care of a VMS system that customers could rent time on, and doing media
conversion for customers. (Lots of weird requests for weird media came
through, and I had to deal with all kind of trouble generating hardware...)
I didn't have a permanent position, and I have no recollection if I had
a badge with a number, or what that number was.
Anyway, it was interesting.
Johnny
On 2023-03-12 21:05, Thomas DeBellis wrote:
Gee, I thought he was asking me...
For what it's worth, I was with DEC in the Galaxy group from late 1978
to late 1980. I worked on the File Finder project, which was an online
database to keep track of backup tapes and the contents therein.
So if you zorched a file and wanted to restore it, you'd run File Finder
and find out what tape (or tapes) it was on and away you went. An
earlier version had been written in COBOL, which was using fixed length
records that were scanned sequentially. It rapidly blew up disk space
as well as having order O(N) CPU usage.
We got as far as designing a b-tree structured database, partly because
of the discovery process involved but mostly because we were given some
very poor implementation restrictions. It's too bad; it could have
easily been integrated into Galaxy, so once you found what you wanted,
you could get the mount request in and trigger the archive system's
restore process, which would have been very seamless. These days,that
kind of functionality is found everywhere (Veritas, Time Machine, Etc.)
I can't remember my badge number, but I still have it. Somewhere... I
can' barely remember what building I was in Marlboro. I think #3? (the
one on the right hand side closest to manufacturing and the upper
parking lot). I do remember the Galaxy group's machine, 2116 ("The Big
Orange Welcomes You") and, oddly enough, my sign-on id, "DEBELLIS" :-)
--
Johnny Billquist || "I'm on a bus
|| on a psychedelic trip
email: bqt(a)softjar.se || Reading murder books
pdp is alive! || tryin' to stay hip" - B. Idol
------------------------------------------------------------------------
On 3/12/23 10:56 AM, Robert Armstrong wrote:
I quit in 1985. Badge number 147367 (still have it memorized :)
Bob
> ------------------------------------------------------------------------
> Johnny Billquist<bqt(a)softjar.se> wrote:
> When were you at DEC?
_______________________________________________
HECnet mailing list -- hecnet(a)lists.dfupdate.se
To unsubscribe send an email to hecnet-leave(a)lists.dfupdate.se
9:24 p.m.
It's exactly similar; in other words, it's the same thing. "archive" is
the Tops-20 term for offline files that are voluntarily put on tape.
The directory entries still exist, but don't point to any data. The FIle
Descriptor Block (FDB) contains two entries for the names of two tapes
with copies of the data. You can set things up so that if you try to
open the file, you get put to sleep until the operator mounts the tape
and then DUMPER goes and finds your file. Seamless, but slow unless you
happened to already have the tape loaded (and it was labeled ANSI or
Tops-20).
Files could be set for forced archiving, known as "migration". The
installation would set expiration dates for files so that if they hadn't
been used in certain period of time, they were "migrated" to tape.
DEC Marlboro was an extremely cool place to be before the Jupiter was
cancelled. Very interesting stuff happening, some of which made its to VMS.
At least in the United States, if you were full time, you had a blue
badge and if you weren't (like a contractor) you got purple. You didn't
get your face on the purple badge.
------------------------------------------------------------------------
On 3/12/23 4:11 PM, Johnny Billquist wrote:
I was actually asking Bob, but I didn't know you were also at DEC. Cool.
What you describe sounds very similar to what offline files in TOPS-20
ended up like.
9:32 p.m.
On 2023-03-12 21:24, Thomas DeBellis wrote:
It's exactly similar; in other words, it's the
same thing. "archive" is
the Tops-20 term for offline files that are voluntarily put on tape.
The directory entries still exist, but don't point to any data. The FIle
Descriptor Block (FDB) contains two entries for the names of two tapes
with copies of the data. You can set things up so that if you tryto
open the file, you get put to sleep until the operator mounts the tape
and then DUMPER goes and finds your file. Seamless, but slow unless you
happened to already have the tape loaded (and it was labeled ANSI or
Tops-20).
Oh, I know. I was managing two TOPS-20 systems at the university, and
sometimes got mount requests for files to be brought back on disk. So
off to the vault and fetch the requested tape, mount it, tell the system
to proceed. I think it was very neat.
Files could be set for forced archiving, known as
"migration". The
installation would set expiration dates for files so that if they hadn't
been used in certain period of time, they were "migrated" to tape.
Yeah. We used that too.
DEC Marlboro was an extremely cool place to be before
the Jupiter was
cancelled. Very interesting stuff happening, some of which made its to VMS.
At least in the United States, if you were full time, you had a blue
badge and if you weren't (like a contractor) you got purple. You didn't
get your face on the purple badge.
I wasn't a contractor either. I was still a student, but for various
reasons got employed in parallel to being a student. Gymnasium in
Sweden. I guess somewhat similar to highschool in the US. So it was more
like a temporary employment, but not with a defined time limit either.
Or at least that is how I remember it.
As far as I can remember, my badge was white. Can't remember if I had a
picture on it or not. But I think all other people around me also had
white badges.
Johnny
--
Johnny Billquist || "I'm on a bus
|| on a psychedelic trip
email: bqt(a)softjar.se || Reading murder books
pdp is alive! || tryin' to stay hip" - B. Idol
8 Mar
8 Mar
11:13 p.m.
I have really enjoyed the 'background' sense of community having several DECnet
nodes connected together here locally with connections across the planet. Compared to
where I came from in DECnet terms - central university collection of three VAX 11-780s
interconnected by a mix of CI and kmv11(?) X25 circuits with a few microVAXes dotted about
the campus connected over serial lines with async DDCMP until Ethernet became a campus
universal, then a mix of Ethernet LAN segments and X25(1984) over Ethernet using PSI
software - HECnet is a wonderful achievement. I'm not the only one who tried, but it
would have been great to get LCM+L's DEC systems connected, Rosie in particular. I
have also cajoled the 'twenex' people and also those in (UK) National Museum of
Computing without success.
Regards
Keith
-----Original Message-----
From: Johnny Billquist [mailto:bqt@softjar.se]
Sent: 08 March 2023 20:47
To: 'The Hobbyist DECnet mailing list' <hecnet(a)lists.dfupdate.se>
Subject: [HECnet] A milestone
I have been planning for a long time to post a small reflection when I reached 1000 nodes
registered in the hecnet nodename database.
This happened tonight. I think it is a pretty cool thing. There are now
1000 nodenames registered in this small hobby DECnet. I guess you could say it's
actually not that small.
However, I know that rather few machines are actually online, and it might even be that
the majority of nodes registered have never been online. My guess is that maybe 5% of the
registered machines are usually online. But I do not have any hard data to back this up.
HECnet started out about 20 years ago from a desire I had to hook up a
PDP-11 I had at home, to some machines at my university, using DECnet.
At the time, I didn't have any TCP/IP for RSX, and the only way to get any kind of
networking was to try and come up with some way of getting DECnet up.
My first implementation basically just forwarded a serial port communication between two
Unix machines. And on each end I then hooked that serial port into a PDP-11 running RSX,
and used DDCMP for the actual DECnet link.
This worked, but was obviously not that fast, as the serial ports were limited to 9600
bps.
After a year or so, I figured I could instead write a small program that would forward
ethernet packets. Using UDP I basically had the same property as a local ethernet, but it
could be located somewhere pretty far away. From a DECnet point of view, it would appear
as if they were on the same ethernet segment, while in reality they were nowhere near.
As DECnet have pretty long timeouts on things, it turned out this worked without a hitch,
and I could achieve much better throughput.
Not long after that, the first other users were hooked up to HECnet as well. This was
maybe around 2003 or so.
Another data point is that there are 113 different persons that have one or more nodenames
registered.
Happy milestone, everyone, and thanks for being around.
Johnny
--
Johnny Billquist || "I'm on a bus
|| on a psychedelic trip
email: bqt(a)softjar.se || Reading murder books
pdp is alive! || tryin' to stay hip" - B. Idol
_______________________________________________
HECnet mailing list -- hecnet(a)lists.dfupdate.se To unsubscribe send an email to
hecnet-leave(a)lists.dfupdate.se
12 Mar
12 Mar
3:33 a.m.
On 3/8/23 5:13 PM, Keith Halewood wrote:
HECnet is a wonderful achievement. I'm not the only one who tried, but it would have
been great to get LCM+L's DEC systems connected, Rosie in particular. I have also
cajoled the 'twenex' people and also those in (UK) National Museum of Computing
without success.
What 'twenex' people would these be? Mark Crispin kind of held that
community together and since he's passed, I don't believe I've had the
opportunity to encounter many former systems programmers, which kind of
surprises me.
Getting a 20 on HECnet is easy enough (forgetting what "easy" might mean
for the moment), but their care and feeding is another matter entirely.
The default software assumes a trained staff and has to be configured to
lessen that dependence. Again, not hairy, just batch jobs and ...
Well, I have yet to tot up the number of changes I've needed to make
and, thus far, they have been substantial.
There was also a fair amount of antipathy in that nobody would seem to
care to understand why you would want DECnet if you had TCP/IP. So
finding a 20 hacker who can do DECnet was fairly hit and miss. I
certainly would not have been able to chase down some of the problems
that I have without Johnny's prodigious talents as well as other people
here familiar with the protocol.
Although MRC did know some of DECnet, he didn't know it like he knew
other areas. I don't recall him publishing a DECnet fix on
Tops-20-Wizards that he himself had done. He certainly did take and
publish changes as well as adding DECnet transport for the SMTP system.
Tops-10 hackers are even more rare as TCP/IP was never a DEC product;
people apparently wrote their own IP stacks, as far as I am aware.
Apparently not all 36 bit hardware will do DECnet. I can't remember
exactly, but I believe the network interfaces on some Toad's can not
have their MAC addresses programmatically changed (anybody know this?)
I do know that one developer at XKL told me that DECnet was no longer in
their monitor and he didn't know if they could even assemble it
anymore. I was surprised at that, but didn't think to follow up at the
time.
And if you're going to run actual KL hardware... Well, God bless you,
but do bring your wallet because you're going to need it, if for nothing
else then the electricity, even if you are using CompuServe's power
supplies. And the hardware was gigantic wire-wrap backplanes which
could sometimes be mysteriously cantankerous for days for no readily
apparent reason although it was usually reliable enough for the times
It's possible that these museums don't have adequate volunteer staff to
keep up with those kinds of things. Maybe I'll volunteer after I get
PANDA II pushed out. I'm just writing up the release notes for
Kermit-20/NRT, so I want to get back to winding the rest of my DECnet
fixes up and then put something out.
13 Mar
13 Mar
2:44 p.m.
On Mar 11, 2023, at 9:33 PM, Thomas DeBellis
<tommytimesharing(a)gmail.com> wrote:
It would be interesting to confirm that. If it's true, then there is a potential
answer: implement Phase IV Prime. That's a thingly documented extension of Phase IV
originally done to support IEEE 802.5 token ring (with its inaccurately named
"functional addresses"). Among other things it supports using the
"hardware address" rather than the Bernie Lacroute "hiord" address.
I'm planning to implement that at some point in PyDECnet to allow it to talk to DECnet
nodes on WiFi, where changing the address doesn't normally work. But if there is any
other hardware with that limitation it would work there as well. The changes are pretty
modest, especially for end nodes.
The protocol details are in the 802.5 datalink architecture spec (EK-DNAP4-TR-001), of all
places. I forgot where I found that document online, the filename is
"trn_dlspec_v100.pdf".
paul
On 3/8/23 5:13 PM, Keith Halewood wrote:
HECnet is a wonderful achievement. I'm not the only one who tried, but it would have
been great to get LCM+L's DEC systems connected, Rosie in particular. I have also
cajoled the 'twenex' people and also those in (UK) National Museum of Computing
without success.
...
Apparently not all 36 bit hardware will do DECnet. I can't remember exactly, but I
believe the network interfaces on some Toad's can not have their MAC addresses
programmatically changed (anybody know this?)
3:14 p.m.
On Mar 13, 2023, at 6:44 AM, Paul Koning
<paulkoning(a)comcast.net> wrote:
I have this document at vahaven.com -
http://www.vaxhaven.com/images/b/ba/EK-DNAP4-TR-001.pdf
Ian
On Mar 11, 2023, at 9:33 PM, Thomas DeBellis
<tommytimesharing(a)gmail.com> wrote:
It would be interesting to confirm that. If it's true, then there is a potential
answer: implement Phase IV Prime. That's a thingly documented extension of Phase IV
originally done to support IEEE 802.5 token ring (with its inaccurately named
"functional addresses"). Among other things it supports using the
"hardware address" rather than the Bernie Lacroute "hiord" address.
I'm planning to implement that at some point in PyDECnet to allow it to talk to
DECnet nodes on WiFi, where changing the address doesn't normally work. But if there
is any other hardware with that limitation it would work there as well. The changes are
pretty modest, especially for end nodes.
The protocol details are in the 802.5 datalink architecture spec (EK-DNAP4-TR-001), of
all places. I forgot where I found that document online, the filename is
"trn_dlspec_v100.pdf".
On 3/8/23 5:13 PM, Keith Halewood wrote:
HECnet is a wonderful achievement. I'm not the only one who tried, but it would have
been great to get LCM+L's DEC systems connected, Rosie in particular. I have also
cajoled the 'twenex' people and also those in (UK) National Museum of Computing
without success.
...
Apparently not all 36 bit hardware will do DECnet. I can't remember exactly, but I
believe the network interfaces on some Toad's can not have their MAC addresses
programmatically changed (anybody know this?)
3:55 p.m.
On 2023-03-13 14:44, Paul Koning wrote:
As far as I know, Thomas is right. The XKL machines cannot speak DECnet.
However, I don't know if it is just the MAC address that is the issue,
or if there was something else/more. I think I talked about this with
both Peter Lothberg and Rich Alderson in the past. One of them should
know the details... If I got things right, there are some more
differences between the XKL and the SC machines as well, when it comes
to memory addressing. But again, I don't remember the details. And SC
machines do speak DECnet.
The problem with Phase IV Prime is that it will not be compatible with
any other Phase IV node, so it would be of very limited value.
On Mar 11, 2023, at 9:33 PM, Thomas DeBellis
<tommytimesharing(a)gmail.com> wrote:
It would be interesting to confirm that. If it's true, then there is a potential
answer: implement Phase IV Prime. That's a thingly documented extension of Phase IV
originally done to support IEEE 802.5 token ring (with its inaccurately named
"functional addresses"). Among other things it supports using the
"hardware address" rather than the Bernie Lacroute "hiord" address.
On 3/8/23 5:13 PM, Keith Halewood wrote:
HECnet is a wonderful achievement. I'm not the only one who tried, but it would have
been great to get LCM+L's DEC systems connected, Rosie in particular. I have also
cajoled the 'twenex' people and also those in (UK) National Museum of Computing
without success.
...
Apparently not all 36 bit hardware will do DECnet. I can't remember exactly, but I
believe the network interfaces on some Toad's can not have their MAC addresses
programmatically changed (anybody know this?) I'm planning to implement that at some point in
PyDECnet to allow it to talk to DECnet nodes on WiFi, where changing the address
doesn't normally work. But if there is any other hardware with that limitation it
would work there as well. The changes are pretty modest, especially for end nodes.
True. But you will only be able to talk to other Phase IV prime nodes.
And I don't think I've ever seen any, so that would boil down to just be
able to talk DECnet to other PyDECnet nodes. Not sure how much value
that would have...
Johnny
--
Johnny Billquist || "I'm on a bus
|| on a psychedelic trip
email: bqt(a)softjar.se || Reading murder books
pdp is alive! || tryin' to stay hip" - B. Idol
4:06 p.m.
On Mar 13, 2023, at 10:55 AM, Johnny Billquist
<bqt(a)softjar.se> wrote:
...
The problem with Phase IV Prime is that it will not be compatible with any other Phase IV
node, so it would be of very limited value.
Phase IV Prime can only talk to prime neighbors, but via a prime router it can talk to
anything else just like regular Phase IV. So it should be useful; the only limitation is
that you'd need to have a Phase IV prime router adjacency.
paul
I'm planning to implement that at some point
in PyDECnet to allow it to talk to DECnet nodes on WiFi, where changing the address
doesn't normally work. But if there is any other hardware with that limitation it
would work there as well. The changes are pretty modest, especially for end nodes.
True. But you will only be able to talk to other Phase IV prime nodes. And I don't
think I've ever seen any, so that would boil down to just be able to talk DECnet to
other PyDECnet nodes. Not sure how much value that would have...
4:55 p.m.
On 2023-03-13 16:06, Paul Koning wrote:
Right. And since the only Prime implementation that I am aware of would
be the PyDECnet one, it means you can talk DECnet directly between
those, but not to anything else. However, you can also achieve this by
just having GRE, Multinet, or whatever in between those two nodes, and
just use IP.
So I'm not sure how much value it could be considered to be to implement
Prime in PyDECnet.
Johnny
--
Johnny Billquist || "I'm on a bus
|| on a psychedelic trip
email: bqt(a)softjar.se || Reading murder books
pdp is alive! || tryin' to stay hip" - B. Idol
On Mar 13, 2023, at 10:55 AM, Johnny Billquist
<bqt(a)softjar.se> wrote:
...
The problem with Phase IV Prime is that it will not be compatible with any other Phase IV
node, so it would be of very limited value.
Phase IV Prime can only talk to prime neighbors, but via a prime router it can talk to
anything else just like regular Phase IV. So it should be useful; the only limitation is
that you'd need to have a Phase IV prime router adjacency. I'm planning to implement that at some point
in PyDECnet to allow it to talk to DECnet nodes on WiFi, where changing the address
doesn't normally work. But if there is any other hardware with that limitation it
would work there as well. The changes are pretty modest, especially for end nodes.
True. But you will only be able to talk to other Phase IV prime nodes. And I don't
think I've ever seen any, so that would boil down to just be able to talk DECnet to
other PyDECnet nodes. Not sure how much value that would have...
6:03 p.m.
On Mar 13, 2023, at 11:55 AM, Johnny Billquist
<bqt(a)softjar.se> wrote:
On 2023-03-13 16:06, Paul Koning wrote:
Right. And since the only Prime implementation that I am aware of would be the PyDECnet
one, it means you can talk DECnet directly between those, but not to anything else.
However, you can also achieve this by just having GRE, Multinet, or whatever in between
those two nodes, and just use IP.
Yes, IP would work for nodes having Multinet support. Can a Toad do that?
But I don't understand your other point. If I have a prime endnode, adjacent to a
prime ("bilingual") router, it can talk to MIM, or anyone else on the HECnet.
There is of course also the point of "I can do it so why not?"
paul
On Mar
13, 2023, at 10:55 AM, Johnny Billquist <bqt(a)softjar.se> wrote:
...
The problem with Phase IV Prime is that it will not be compatible with any other Phase IV
node, so it would be of very limited value.
Phase IV Prime can
only talk to prime neighbors, but via a prime router it can talk to anything else just
like regular Phase IV. So it should be useful; the only limitation is that you'd need
to have a Phase IV prime router adjacency. I'm planning to implement that at some point
in PyDECnet to allow it to talk to DECnet nodes on WiFi, where changing the address
doesn't normally work. But if there is any other hardware with that limitation it
would work there as well. The changes are pretty modest, especially for end nodes.
True. But you will only be able to talk to other Phase IV prime nodes. And I don't
think I've ever seen any, so that would boil down to just be able to talk DECnet to
other PyDECnet nodes. Not sure how much value that would have...
6:07 p.m.
On 2023-03-13 18:03, Paul Koning wrote:
No. A Toad cannot talk DECnet at all, as far as I am aware. It also do
not have Prime support, so you would not be able to talk DECnet to it
even if you did implement this. Basically, for the time being, Toads are
out of DECnet, in all aspects. They are IP only.
On Mar 13, 2023, at 11:55 AM, Johnny Billquist
<bqt(a)softjar.se> wrote:
On 2023-03-13 16:06, Paul Koning wrote:
Right. And since the only Prime implementation that I am aware of would be the PyDECnet
one, it means you can talk DECnet directly between those, but not to anything else.
However, you can also achieve this by just having GRE, Multinet, or whatever in between
those two nodes, and just use IP.
Yes, IP would work for nodes having Multinet support. Can a Toad do that? On Mar
13, 2023, at 10:55 AM, Johnny Billquist <bqt(a)softjar.se> wrote:
...
The problem with Phase IV Prime is that it will not be compatible with any other Phase IV
node, so it would be of very limited value.
> I'm planning to implement that at some point in PyDECnet to allow it to talk to
DECnet nodes on WiFi, where changing the address doesn't normally work. But if there
is any other hardware with that limitation it would work there as well. The changes are
pretty modest, especially for end nodes.
True. But you will only be able to talk to other Phase IV prime nodes. And I don't
think I've ever seen any, so that would boil down to just be able to talk DECnet to
other PyDECnet nodes. Not sure how much value that would have...
Phase IV Prime can
only talk to prime neighbors, but via a prime router it can talk to anything else just
like regular Phase IV. So it should be useful; the only limitation is that you'd need
to have a Phase IV prime router adjacency. But I don't understand your other point. If I
have a prime endnode, adjacent to a prime ("bilingual") router, it can talk to
MIM, or anyone else on the HECnet.
Correct. But my point is still that there is noone talking Prime. Your
implementation would be the only one, so you could only use it to talk
to other PyDECnet nodes. And those PyDECnet nodes can already talk to
each other, even over WiFi. So I see very limited value in implementing
Prime.
There is of course also the point of "I can do it
so why not?"
Always a valid point. :-)
Johnny
--
Johnny Billquist || "I'm on a bus
|| on a psychedelic trip
email: bqt(a)softjar.se || Reading murder books
pdp is alive! || tryin' to stay hip" - B. Idol
6:47 p.m.
On Mar 13, 2023, at 1:07 PM, Johnny Billquist
<bqt(a)softjar.se> wrote:
On 2023-03-13 18:03, Paul Koning wrote:
No. A Toad cannot talk DECnet at all, as far as I am aware. It also do not have Prime
support, so you would not be able to talk DECnet to it even if you did implement this.
Basically, for the time being, Toads are out of DECnet, in all aspects. They are IP only.
My thinking was that a Toad runs TOPS-20 so it should be able to run DECnet, except for
some details. If a fixed MAC address stands in the way but someone feels like tweaking
DECnet/20 to add Prime support, that would be a way to cure it.
If the difficulties for DECnet/Toad are much larger than that, I suppose this doesn't
really help.
paul
On Mar
13, 2023, at 11:55 AM, Johnny Billquist <bqt(a)softjar.se> wrote:
On 2023-03-13 16:06, Paul Koning wrote:
Yes, IP would work for nodes having Multinet
support. Can a Toad do that? > On Mar 13, 2023, at 10:55 AM, Johnny
Billquist <bqt(a)softjar.se> wrote:
>
> ...
> The problem with Phase IV Prime is that it will not be compatible with any other
Phase IV node, so it would be of very limited value.
>
>> I'm planning to implement that at some point in PyDECnet to allow it to talk
to DECnet nodes on WiFi, where changing the address doesn't normally work. But if
there is any other hardware with that limitation it would work there as well. The changes
are pretty modest, especially for end nodes.
>
> True. But you will only be able to talk to other Phase IV prime nodes. And I
don't think I've ever seen any, so that would boil down to just be able to talk
DECnet to other PyDECnet nodes. Not sure how much value that would have...
Phase IV Prime can only talk to prime neighbors, but via a prime router it can talk to
anything else just like regular Phase IV. So it should be useful; the only limitation is
that you'd need to have a Phase IV prime router adjacency.
Right. And since the only Prime implementation that I am aware of would be the PyDECnet
one, it means you can talk DECnet directly between those, but not to anything else.
However, you can also achieve this by just having GRE, Multinet, or whatever in between
those two nodes, and just use IP. 
9 Mar
9 Mar
12:29 a.m.
Johnny,
Congrats on achieving One Thousand HECnet nodes!
Your email got me interested in when I first started using your TCP/IP stack for
RSX11M+. It looks like that happened around March of 2013. Our first conversations were in
2011 around Datatrieve-11 and its call interface. In those early days, we also talked
about recovering/fixing APL-11 (which you did a few years ago). We also talked about the
Saturn software for which I am still trying to get the sources.
It is amazing how far you’ve taken the TCP/IP and HECnet network! Thanks for all the
many, many hours of work (and hopefully some fun for you) you’ve given us!!
Thanks!!!
Mark
On Mar 8, 2023, at 2:47 PM, Johnny Billquist
<bqt(a)softjar.se> wrote:
I have been planning for a long time to post a small reflection when I reached 1000 nodes
registered in the hecnet nodename database.
This happened tonight. I think it is a pretty cool thing. There are now 1000 nodenames
registered in this small hobby DECnet. I guess you could say it's actually not that
small.
However, I know that rather few machines are actually online, and it might even be that
the majority of nodes registered have never been online. My guess is that maybe 5% of the
registered machines are usually online. But I do not have any hard data to back this up.
HECnet started out about 20 years ago from a desire I had to hook up a PDP-11 I had at
home, to some machines at my university, using DECnet. At the time, I didn't have any
TCP/IP for RSX, and the only way to get any kind of networking was to try and come up with
some way of getting DECnet up.
My first implementation basically just forwarded a serial port communication between two
Unix machines. And on each end I then hooked that serial port into a PDP-11 running RSX,
and used DDCMP for the actual DECnet link.
This worked, but was obviously not that fast, as the serial ports were limited to 9600
bps.
After a year or so, I figured I could instead write a small program that would forward
ethernet packets. Using UDP I basically had the same property as a local ethernet, but it
could be located somewhere pretty far away. From a DECnet point of view, it would appear
as if they were on the same ethernet segment, while in reality they were nowhere near. As
DECnet have pretty long timeouts on things, it turned out this worked without a hitch, and
I could achieve much better throughput.
Not long after that, the first other users were hooked up to HECnet as well. This was
maybe around 2003 or so.
Another data point is that there are 113 different persons that have one or more
nodenames registered.
Happy milestone, everyone, and thanks for being around.
Johnny
--
Johnny Billquist || "I'm on a bus
|| on a psychedelic trip
email: bqt(a)softjar.se || Reading murder books
pdp is alive! || tryin' to stay hip" - B. Idol
_______________________________________________
HECnet mailing list -- hecnet(a)lists.dfupdate.se
To unsubscribe send an email to hecnet-leave(a)lists.dfupdate.se
12:41 a.m.
I saw 999 nodes this morning when I updated my database - somebody else
must have signed up today :)
Pretty neat, but as you say I wonder how many names are just reserved for
future expansion, and how many really exist?
And I wonder how many are real, physical, machines and how many are
simulations?
And how many are real machines, but not actual DEC machines? (yes, there
are some!)
It'd be very interesting to see some statistics...
Bob
12:59 a.m.
I normally have two PDP-11/83s, a VAXstation 4000/90A, MicroVAX 3100-80 up on HECnet
247/365. Then on occasion and XP1000 Alpha, and a PDP-11/84. On the non-DEC side, there
are also a couple PiDP-11/70s also running on HECnet 24/7/365. Also in the lab is a
PDP-11/05 and a MINC-23 but they can’t do DECnet although they do have file transfer
capability via their UniBone and Qbone interfaces.
On Mar 8, 2023, at 5:41 PM, Robert Armstrong
<bob(a)jfcl.com> wrote:
I saw 999 nodes this morning when I updated my database - somebody else
must have signed up today :)
Pretty neat, but as you say I wonder how many names are just reserved for
future expansion, and how many really exist?
And I wonder how many are real, physical, machines and how many are
simulations?
And how many are real machines, but not actual DEC machines? (yes, there
are some!)
It'd be very interesting to see some statistics...
Bob
_______________________________________________
HECnet mailing list -- hecnet(a)lists.dfupdate.se
To unsubscribe send an email to hecnet-leave(a)lists.dfupdate.se
1:24 a.m.
On 2023-03-09 00:41, Robert Armstrong wrote:
I saw 999 nodes this morning when I updated my
database - somebody else
must have signed up today :)
Yup. There was. ;-)
Pretty neat, but as you say I wonder how many names
are just reserved for
future expansion, and how many really exist?
Some are future expansion, but I think an even larger number is just
machines that is sitting somewhere, or disk images of systems, which
*might* get activated at some point.
Almost every nodename does also have associated architecture, CPU and OS
written down, which is more than just a placeholder, I'd say.
And I wonder how many are real, physical, machines
and how many are
simulations?
That is a good question. I have sortof some slight notion of this, but
not perfect.
DTR> find hecnet with cpu cont "simh"
[265 records found]
Which of course do not count other simulations, nor is every one marked
as simh.
So here are some more "fun" statistics:
DTR> sum 1 by architecture
ARCHITECTURE
? 357
ALPHA 102
H6000 2
i386 46
IA64 16
MIPS 2
PDP10 43
PDP11 118
POWERPC 2
VAX 312
1000
DTR> sum 1 by os
OS
? 310
CentOS 7 1
DECnet DOS 2
Digital Unix V4.0E 1
Digital Unix V4.0G 1
FastPath 4 1
FreeBSD-12 1
IOS 5
IOS V10.3 1
IOS V12 1
IOS V12.1 1
IOS V12.2 1
IOS V12.3 3
IOS V12.4 3
Linux 35
MacOS 3
MacOS 9 1
MicroVMS V4.6 1
Multics MR12.6f 2
OpenVMS 20
OpenVMS V6.0 1
OpenVMS V6.1 8
OpenVMS V6.2 3
OpenVMS V7.0 1
OpenVMS V7.1 2
OpenVMS V7.2 14
OpenVMS V7.2-1 2
OpenVMS V7.3 208
OpenVMS V7.3-1 9
OpenVMS V7.3-2 8
OpenVMS V8.3 24
OpenVMS V8.3-1H1 1
OpenVMS V8.4 30
OpenVMS V8.4-2L1 3
OpenVMS V8.4-2L3 1
OpenVMS V8.4E 1
OSF/1 1
P/OS 1
P/OS V3.2 2
PyDECnet 26
RSTS/E 2
RSTS/E V10 2
RSTS/E V10.0 1
RSTS/E V10.1 13
RSTS/E V10.1-A 1
RSTS/E V10.1-L 2
RSTS/E V9.6 1
RSX 1
RSX or RSTS/E 1
RSX-11M 1
RSX-11M V4.2 2
RSX-11M V4.4 1
RSX-11M V4.6 1
RSX-11M V4.8 3
RSX-11M-PLUS 6
RSX-11M-PLUS V4.2 1
RSX-11M-PLUS V4.6 66
RSX-11M-PLUS V5.0 1
RT-11 1
RT-11 V5.2 1
System 7 1
Tops-10 7
Tops-10 V7.04 2
Tops-10 V7.05 1
TOPS-20 4
TOPS-20 (Panda) 5
TOPS-20 V4.1 3
TOPS-20 V7.0 12
TOPS-20 V7.1 4
Tru64 2
Tru64 V5.0 2
Tru64 V5.1 1
Tru64 V5.1a 2
Tru64 V5.1B 1
Tru64 V5.1B6 1
Ultrix 3
Ultrix V4.0 6
Ultrix V4.2 1
Ultrix V4.3 2
Ultrix V4.4 1
Ultrix V4.5 6
VAX/VMS 2
VAX/VMS V2,0 1
VAX/VMS V3.5 1
VAX/VMS V4.0 1
VAX/VMS V4.4 1
VAX/VMS V4.7 8
VAX/VMS V5.0 1
VAX/VMS V5.1 1
VAX/VMS V5.2 1
VAX/VMS V5.3 2
VAX/VMS V5.4 6
VAX/VMS V5.4-3 1
VAX/VMS V5.5 3
VAX/VMS V5.5-1 3
VAX/VMS V5.5-2 6
VAX/VMS V5.5-2H4 2
VAX/VMS V5.5-5 1
VAX/VMS V6.0 1
VAX/VMS V6.1 6
VAX/VMS V6.2 3
VAXELN V4.6 2
VMS 6
VMS or Ultrix 1
VMS/Alpha V1.5-1H1 1
VMScluster 5
VSI VMS 1
Windows 2000 Srv 1
Windows 2003 1
Windows 3.1 1
Windows 3.11 1
Windows NT 4
Windows NT V4.0 1
Windows Server 2003 1
Windows XP 8
1000
And how many are real machines, but not actual DEC
machines? (yes, there
are some!)
It'd be very interesting to see some statistics...
I think the price for the oddest systems might be Multics.
My statistics are, as mentioned, not perfect, as the nodename database
is not always fully correct or even complete. But it do give some fun
statistics at least.
Pretty fun, the stuff I can easily do from Datatrieve with what I have
stored in there.
Johnny
--
Johnny Billquist || "I'm on a bus
|| on a psychedelic trip
email: bqt(a)softjar.se || Reading murder books
pdp is alive! || tryin' to stay hip" - B. Idol
2 a.m.
On 3/8/23 19:48, Robert Armstrong wrote:
I think we should drop Cisco IOS and pyDECnet from
the count - those
aren't really OSes for our purposes.
I agree that they're not, but I think we should keep them in there,
to identify the nodes. Perhaps add a flag in the database schema saying
it's a router, rather than a "true" DECnet node?
-Dave
--
Dave McGuire, AK4HZ
New Kensington, PA
2:10 a.m.
On Mar 8, 2023, at 8:00 PM, Dave McGuire
<mcguire(a)neurotica.com> wrote:
On 3/8/23 19:48, Robert Armstrong wrote:
Um, routers are true DECnet nodes. The defintion of a DECnet node is "a software
construct that implements (a suitable subset of) the DECnet protocols". It has been
true for a very long time that not all DECnet nodes are general purpose computers in which
DECnet is merely one of many OS services. Routers, as dedicated devices, appeared in the
early 1980s. And that isn't even counting PDP-10 front ends running Phase II
Intercept service; those are routers too, though the name wasn't applied at the time.
For that matter, there are also DECnet nodes that are neither routers nor general purpose
computers. The network-connected PostScript printers (LPS40, LPS20) are an example, as
are some of the earlier X terminals. (Wasn't there something called the VT1000?)
paul
I think we should drop Cisco IOS and pyDECnet
from the count - those
aren't really OSes for our purposes.
I agree that they're not, but I think we should keep them in there, to identify the
nodes. Perhaps add a flag in the database schema saying it's a router, rather than a
"true" DECnet node?
-Dave
2:14 a.m.
Um, routers are true DECnet nodes.
They're DECnet nodes, but they aren't "computers". You can't do
anything
with them except route messages. Can't log in, edit files, compile
programs, play Adventure or Star Trek, nothing :)
We have different goals - I'm not counting nodes, I'm counting general
purpose computing hardware.
Bob
2:20 a.m.
On 2023-03-09 02:14, Robert Armstrong wrote:
Depending on the system, that might be true. There are very different
type of nodes...
Um, routers
are true DECnet nodes.
They're DECnet nodes, but they aren't "computers". You can't do
anything
with them except route messages. Can't log in, edit files, compile
programs, play Adventure or Star Trek, nothing :) We have different goals - I'm not counting
nodes, I'm counting general
purpose computing hardware.
Well, the nodename database don't make any such distinction. The purpose
for it, in the end, is the mapping between DECnet node numbers and
names. Which is meaningful for any kind of node.
Johnny
--
Johnny Billquist || "I'm on a bus
|| on a psychedelic trip
email: bqt(a)softjar.se || Reading murder books
pdp is alive! || tryin' to stay hip" - B. Idol
2:32 a.m.
On 3/8/23 20:14, Robert Armstrong wrote:
It really seems a flag is needed. The node database can enumerate
computing hardware, which is of value, but it can also enumerate all
nodes, whatever their purpose or capabilities.
-Dave
--
Dave McGuire, AK4HZ
New Kensington, PA
Um, routers
are true DECnet nodes.
They're DECnet nodes, but they aren't "computers". You can't do
anything
with them except route messages. Can't log in, edit files, compile
programs, play Adventure or Star Trek, nothing :)
We have different goals - I'm not counting nodes, I'm counting general
purpose computing hardware.
2:42 a.m.
On 2023-03-09 02:32, Dave McGuire wrote:
On 3/8/23 20:14, Robert Armstrong wrote:
It really seems a flag is needed. The node database can enumerate
computing hardware, which is of value, but it can also enumerate all
nodes, whatever their purpose or capabilities.
While I can of course add such a flag, at the moment I'm not really
getting why.
Are we just trying to make some kind of distinction between an embedded
system and some generic user environment?
Something like IOS can certainly be used interactively, and you can
manipulate the system and work on various aspects. But in general, I do
not expect that you can run arbitrary binaries on it.
On the other hand, some network booted RSX-11S system with no file
system, what would you classify that as? It can also only run binaries
that are included in the system image, or possibly loaded over the
network. But it;s still a PDP-11. Isn't that "general purpose computing
hardware"?
It's always extremely risky to try and make distinctions based on some
subjective definition which is at best very floating.
I think we'd first need to define what we actually would want to accomplish.
Johnny
--
Johnny Billquist || "I'm on a bus
|| on a psychedelic trip
email: bqt(a)softjar.se || Reading murder books
pdp is alive! || tryin' to stay hip" - B. Idol
Um,
routers are true DECnet nodes.
They're DECnet nodes, but they aren't "computers". You can't do
anything
with them except route messages. Can't log in, edit files, compile
programs, play Adventure or Star Trek, nothing :)
We have different goals - I'm not counting nodes, I'm counting general
purpose computing hardware.
3:06 a.m.
On 3/8/23 20:42, Johnny Billquist wrote:
While I can of course add such a flag, at the moment I'm not really
getting why.
Are we just trying to make some kind of distinction between an embedded
system and some generic user environment?
Hey, I'm not the one who wanted to not have them in the database! ;)
Personally, I think a node database should be exactly that, a node
database. If it has a DECnet address, it should be in the node database.
We're very good at splitting hairs here. ;)
-Dave
--
Dave McGuire, AK4HZ
New Kensington, PA
They're DECnet nodes, but they aren't
"computers". You can't do
anything
with them except route messages. Can't log in, edit files, compile
programs, play Adventure or Star Trek, nothing :)
We have different goals - I'm not counting nodes, I'm counting
general
purpose computing hardware.
It really seems a flag is needed. The node database can enumerate
computing hardware, which is of value, but it can also enumerate all
nodes, whatever their purpose or capabilities.
3:24 a.m.
Johnny Billquist <bqt(a)softjar.se> wrote:
While I can of course add such a flag, at the moment
I'm not really
getting why.
At least for me, I'm not suggesting that you change anything; the OS type/name is
sufficient for my needs. I was just commenting on the total number of
"computers" on HECnet.
And yes, I admit that my definition of "computer" is somewhat arbitrary (hence
the quotation marks!) but I know one when I see one :)
Bob
12 Mar
12 Mar
2:28 a.m.
On 3/8/23 8:14 PM, Robert Armstrong wrote:
They're DECnet nodes, but they aren't "computers". You can't do
anything
with them except route messages. Can't log in, edit files, compile
programs, play Adventure or Star Trek, nothing :)
As far as I am aware, all PANDA Tops-20 distributions have Advent. I
have an Algol version of Star Trek, but I have yet to track the authors
down.
Finding out what a node can do has been done in different ways. Tops-20
had TCP/IP as a DEC product far before other DEC offerings. Not that
you couldn't get it, just not from DEC. If I'm remembering correctly, I
believe one vendor was called Wollongong and maybe it was Australian.
The TCP/IP implementations had earlier and later versions of DNS.
HOSTS.TXT will tell you what services the node offers, but that had to
be kept up to date and it eventually became infeasible. I would imagine
that DNS records could hold other bits of Interest, but I only have what
kind of mail it will do. The Tops-20 DNS (CHIVES) has records for both
CHAOS nodes and IP, this is unremarkable since it comes from MIT. I am
maybe about one fourth of the way through putting in a DECnet record
type. For now, I've modified it to better support DECnet node numbers,
which are not known by default.
I would imagine you could have entries for whether the node allows
public access, anonymous transfer and I guess games.
We have different goals - I'm not counting
nodes, I'm counting general
purpose computing hardware.
I'm actually seeing 1,004 nodes defined.
SETNODE>LIST (node table) totAL (nodes defined in node tables)
Total Unchanged nodes: 1004
Total updates to perform: NONE
Total nodes to delete: NONE
Total words, shared text: 1813
Total nodes mapped: 1004
Section mapped input file:
TOMMYT:<SYSTEM>NODE-DATA.BIN.1;PROCESS-ONLY-JFN:26 ;UNRESTRICTED-READ
Total file pages mapped: 4
I don't know how many nodes are actually up 24/7, but my 20's
(TOMMYT::and VENTI2::) are, viz:
TOMMYT up for 21 Weeks, 6 Days, 21 Hours, 37 Minutes, 8 Seconds and 407
Milliseconds
VENTI2 up for 21 Weeks, 6 Days, 21 Hours, 27 Minutes, 20 Seconds and 773
Milliseconds
9 Mar
9 Mar
2:17 a.m.
On 2023-03-09 02:10, Paul Koning wrote:
> On Mar 8, 2023, at 8:00 PM, Dave McGuire <mcguire(a)neurotica.com> wrote:
>
> On 3/8/23 19:48, Robert Armstrong wrote:
>> I think we should drop Cisco IOS and pyDECnet from the count - those
>> aren't really OSes for our purposes.
>
> I agree that they're not, but I think we should keep them in there, to identify
the nodes. Perhaps add a flag in the database sc
hema saying it's a router,
rather than a "true" DECnet node?
-Dave
4:43 a.m.
And there are the odd beasts like the DECnet/SNA Gateway and the like. Did that code
survive at all?
Groet,
Wilm
(Verstuurd vanaf mijn telefoon, dus wat korter dan gewoonlijk.)
(Sent from my phone, so a bit more compact than usual)
Op 9 mrt. 2023 om 02:17 heeft Johnny Billquist
<bqt(a)softjar.se> het volgende geschreven:
On 2023-03-09 02:10, Paul Koning wrote:
>> On Mar 8, 2023, at 8:00 PM, Dave McGuire
<mcguire(a)neurotica.com> wrote:
>
> On 3/8/23 19:48, Robert Armstrong wrote:
>> I think we should drop Cisco IOS and pyDECnet from the count - those
>> aren't really OSes for our purposes.
>
> I agree that they're not, but I think we should keep them in there, to identify
the nodes. Perhaps add a flag in the database sc
hema saying it's a router,
rather than a "true" DECnet node?
>
> -Dave
_______________________________________________
HECnet mailing list -- hecnet(a)lists.dfupdate.se
To unsubscribe send an email to hecnet-leave(a)lists.dfupdate.se
2:24 a.m.
(Somehow my first response got very garbled...)
On 2023-03-09 02:10, Paul Koning wrote:
All true. And no different than an IP router, for example.
> On Mar 8, 2023, at 8:00 PM, Dave McGuire <mcguire(a)neurotica.com> wrote:
>
> On 3/8/23 19:48, Robert Armstrong wrote:
>> I think we should drop Cisco IOS and pyDECnet from the count - those
>> aren't really OSes for our purposes.
>
> I agree that they're not, but I think we should keep them in there, to identify
the nodes. Perhaps add a flag in the database sc
hema saying it's a router,
rather than a "true" DECnet node?
-Dave
Um, routers are true DECnet nodes. The defintion of a DECnet node is "a software
construct that
implements (a suitable subset of) the DECnet protocols". It has been true for a
very long time
that not all DECnet nodes are general purpose computers in which DECnet is merely one of
many OS services.
Routers, as dedicated devices, appeared in the early 1980s. And that isn't even
counting PDP-10 front ends
running Phase II Intercept service; those are routers too, though the name wasn't
applied at the time. For that matter, there are also DECnet nodes that are
neither routers nor general purpose computers.
The network-connected PostScript printers (LPS40, LPS20) are an example, as are some of
the earlier
X terminals. (Wasn't there something called the VT1000?)
There was both VXT1000 and VXT1200.
Johnny
--
Johnny Billquist || "I'm on a bus
|| on a psychedelic trip
email: bqt(a)softjar.se || Reading murder books
pdp is alive! || tryin' to stay hip" - B. Idol
2:36 a.m.
On 3/8/23 20:24, Johnny Billquist wrote:
And VT1000.
-Dave
--
Dave McGuire, AK4HZ
New Kensington, PA
For that
matter, there are also DECnet nodes that are neither routers
nor general purpose computers.
The network-connected PostScript printers (LPS40, LPS20) are an
example, as are some of the earlier
X terminals. (Wasn't there something called the VT1000?)
There was both VXT1000 and VXT1200.
2:48 a.m.
On 2023-03-09 02:36, Dave McGuire wrote:
On 3/8/23 20:24, Johnny Billquist wrote:
And VT1000.
I was pretty sure they were called VXT and not VT, but googling seems to
suggest they were indeed called VT1000 and VT1200. There was also the
VT1300 and VT2000, which I barely remembered.
Anyway, I was wondering if they talked DECnet, and the Wikipedia article
seems to suggest they actually didn't. They just spoke TCP/IP for X11,
but they also talked LAT natively. But then you just had 24x80 terminals
on that fancy screen. (And of course, they also had serial ports.)
Johnny
--
Johnny Billquist || "I'm on a bus
|| on a psychedelic trip
email: bqt(a)softjar.se || Reading murder books
pdp is alive! || tryin' to stay hip" - B. Idol
For that
matter, there are also DECnet nodes that are neither routers
nor general purpose computers.
The network-connected PostScript printers (LPS40, LPS20) are an
example, as are some of the earlier
X terminals. (Wasn't there something called the VT1000?)
There was both VXT1000 and VXT1200.
2:55 a.m.
On Mar 8, 2023, at 8:48 PM, Johnny Billquist
<bqt(a)softjar.se> wrote:
On 2023-03-09 02:36, Dave McGuire wrote:
I'm not really sure, but at the time I participated in the design review of the first
of those devices DEC wasn't doing TCP much yet, and was very much into X over DECnet.
paul
On 3/8/23 20:24, Johnny Billquist wrote:
And VT1000.
I was pretty sure they were called VXT and not VT, but googling seems to suggest they
were indeed called VT1000 and VT1200. There was also the VT1300 and VT2000, which I barely
remembered.
Anyway, I was wondering if they talked DECnet, and the Wikipedia article seems to suggest
they actually didn't. They just spoke TCP/IP for X11, but they also talked LAT
natively. But then you just had 24x80 terminals on that fancy screen. (And of course, they
also had serial ports.)
Johnny For that
matter, there are also DECnet nodes that are neither routers nor general purpose
computers.
The network-connected PostScript printers (LPS40, LPS20) are an example, as are some of
the earlier
X terminals. (Wasn't there something called the VT1000?)
There was both VXT1000 and VXT1200. 
3:04 a.m.
Definitely had DECnet. The VT1300 was based on VAXELN, and was a stop-gap
until the (much less costly) VT1200 could be finished and put into
production. Based on a diskless VAXstation 3100. Software was done mostly
by the X11 group of the VAXELN group. Still have a system nameplate around
here someplace.
Chuck
On Wed, Mar 8, 2023 at 8:55 PM Paul Koning <paulkoning(a)comcast.net> wrote:
I was pretty sure they were called VXT and not VT, but googling seems to suggest
they were indeed called VT1000 and VT1200. There was also the
VT1300 and VT2000, which I barely remembered.
On Mar 8, 2023, at 8:48 PM, Johnny Billquist
<bqt(a)softjar.se> wrote:
On 2023-03-09 02:36, Dave McGuire wrote:
> On 3/8/23 20:24, Johnny Billquist wrote:
>>> For that matter, there are also DECnet nodes that are neither routers
nor general purpose computers.
>>> The network-connected PostScript
printers (LPS40, LPS20) are an
example, as are some of the earlier
> X terminals. (Wasn't there something called
the VT1000?)
There was both VXT1000 and VXT1200.
And VT1000.
Anyway, I was wondering if they talked DECnet, and the Wikipedia article
seems to
suggest they actually didn't. They just spoke TCP/IP for X11, but
they also talked LAT natively. But then you just had 24x80 terminals on
that fancy screen. (And of course, they also had serial ports.)
Johnny
I'm not really sure, but at the time I participated in the design review
of the first of those devices DEC wasn't doing TCP much yet, and was very
much into X over DECnet.
paul
_______________________________________________
HECnet mailing list -- hecnet(a)lists.dfupdate.se
To unsubscribe send an email to hecnet-leave(a)lists.dfupdate.se
3:04 a.m.
On 3/8/23 20:48, Johnny Billquist wrote:
And VT1000.
I was pretty sure they were called VXT and not VT, but googling seems to
suggest they were indeed called VT1000 and VT1200. There was also the
VT1300 and VT2000, which I barely remembered.
There were also VXT* models.
For that
matter, there are also DECnet nodes that are neither
routers nor general purpose computers.
The network-connected PostScript printers (LPS40, LPS20) are an
example, as are some of the earlier
X terminals. (Wasn't there something called the VT1000?)
There was both VXT1000 and VXT1200. Anyway, I was wondering if they talked DECnet, and the
Wikipedia article
seems to suggest they actually didn't. They just spoke TCP/IP for X11,
but they also talked LAT natively. But then you just had 24x80 terminals
on that fancy screen. (And of course, they also had serial ports.)
I never tried using DECnet, but I sat in front of a VT1000 every day
for a long time, back-ended by a VAX. It was configured for IP.
-Dave
--
Dave McGuire, AK4HZ
New Kensington, PA
2:35 a.m.
On 3/8/23 20:10, Paul Koning wrote:
Um, routers are true DECnet nodes.
I meant "a general-purpose computer that speaks DECnet". (thought
that was clear..)
-Dave
--
Dave McGuire, AK4HZ
New Kensington, PA
I think we should drop Cisco IOS and pyDECnet from
the count - those
aren't really OSes for our purposes.
I agree that they're not, but I think we should keep them in there, to identify the
nodes. Perhaps add a flag in the database schema saying it's a router, rather than a
"true" DECnet node?
2:03 a.m.
On 2023-03-09 01:48, Robert Armstrong wrote:
Two persons. Since the persons have not requested to be anonymous,
anyone can actually find this out.
http://mim.stupi.net/nodedb?search=vaxeln&field=0&sort=0
(Isn't it fun what can be done with RSX, Datatrieve, some CGI and my IP
stack...? :-) )
DTR> sum 1 by os
Who has VAX Elan running? What's that machine doing? I think we should drop Cisco IOS and pyDECnet from
the count - those
aren't really OSes for our purposes.
I struggled back and forth on those. At the moment I've come down on
marking them that way.
For PyDECnet, I think it's pretty meaningless to say "Linux" or
"Windows" or whatever. PyDECnet is like its own standalone DECnet
environment that only exists inside that code Paul have written. Very
different from someone actually running Linux/DECnet for instance.
The even more funny thing about PyDECnet is how I should call the CPU in
this case. I've generally gone for saying the CPU is also PyDECnet, and
architecture is unknown. But I've spotted a few times when I've been
inconsistent.
As for Cisco IOS, why would you not say that this is properly the OS?
In general, if people spot something odd or wrong in the nodename
database, please ping me.
Johnny
--
Johnny Billquist || "I'm on a bus
|| on a psychedelic trip
email: bqt(a)softjar.se || Reading murder books
pdp is alive! || tryin' to stay hip" - B. Idol
2:13 a.m.
On Mar 8, 2023, at 8:03 PM, Johnny Billquist
<bqt(a)softjar.se> wrote:
...
Good point, "Linux" or the like are somewhat misleading. It might make sense to
call the CPU "Python" since what's being executed is Python compiled
bytecode. As for the architecture, either Python or the machine ISP below (like x86 or
Arm).
I think we should drop Cisco IOS and pyDECnet
from the count - those
aren't really OSes for our purposes.
I struggled back and forth on those. At the moment I've come down on marking them
that way.
For PyDECnet, I think it's pretty meaningless to say "Linux" or
"Windows" or whatever. PyDECnet is like its own standalone DECnet environment
that only exists inside that code Paul have written. Very different from someone actually
running Linux/DECnet for instance.
The even more funny thing about PyDECnet is how I should call the CPU in this case.
I've generally gone for saying the CPU is also PyDECnet, and architecture is unknown.
But I've spotted a few times when I've been inconsistent. As for Cisco IOS, why would you not say that this is
properly the OS?
Absolutely. It's an RTOS -- if you can show DECnet nodes running ELAN, it's
equally valid to show DECnet nodes running IOS.
aul
4:39 a.m.
I am running VAXeln on a simh node. It is just sitting there purring and being proud of
itself.
Along with the rest, I am enjoying HECnet so much. Thank you Johnny and all you other
enthusiasts.
Groet,
Wilm
(Verstuurd vanaf mijn telefoon, dus wat korter dan gewoonlijk.)
(Sent from my phone, so a bit more compact than usual)
Op 9 mrt. 2023 om 01:48 heeft Robert Armstrong
<bob(a)jfcl.com> het volgende geschreven:
DTR> sum 1 by os
Who has VAX Elan running? What's that machine doing?
I think we should drop Cisco IOS and pyDECnet from the count - those
aren't really OSes for our purposes.
Bob
_______________________________________________
HECnet mailing list -- hecnet(a)lists.dfupdate.se
To unsubscribe send an email to hecnet-leave(a)lists.dfupdate.se 
11:16 a.m.
Hi,
I hope you don't mind me asking a novice question. I was an RSX-11M guru
rather than a VAX one - I never did much on the new fangled things.
However, when looking to set up a simh VAX the other day, I see that
there are licences required - how would I as a hobbyis tget one? Do I
still have to try to joing DECUS (if it still exists) or what is the
process for requesting a hobbyist licence to bring up a VAX node on HECnet?
Thanks
Peter
On 09/03/2023 03:39, Wilm Boerhout wrote:
I am running VAXeln on a simh node. It is just sitting
there purring and being proud of itself.
Along with the rest, I am enjoying HECnet so much. Thank you Johnny and all you other
enthusiasts.
Groet,
Wilm
(Verstuurd vanaf mijn telefoon, dus wat korter dan gewoonlijk.)
(Sent from my phone, so a bit more compact than usual)
Op 9 mrt. 2023 om 01:48 heeft Robert Armstrong
<bob(a)jfcl.com> het volgende geschreven:
_______________________________________________
HECnet mailing list -- hecnet(a)lists.dfupdate.se
To unsubscribe send an email to hecnet-leave(a)lists.dfupdate.se DTR> sum 1 by os
Who has VAX Elan
running? What's that machine doing?
I think we should drop Cisco IOS and pyDECnet from the count - those
aren't really OSes for our purposes.
Bob
_______________________________________________
HECnet mailing list -- hecnet(a)lists.dfupdate.se
To unsubscribe send an email to hecnet-leave(a)lists.dfupdate.se
11:19 a.m.
Peter,
The VAX Hobbyist license program has been discontinued by the new owners (VSI) since
somewhere last year.
The only practical (legal?) option is now to run VMS V4.6, that precedes the PAK license
program.
I am not going to tell you about the VMS Liberation Front, because that would be against
the law.
Wilm
-----Original Message-----
From: Peter Whisker <peter(a)whisker.org.uk>
Sent: Thursday, March 9, 2023 11:16 AM
To: hecnet(a)lists.dfupdate.se
Subject: [HECnet] Re: A milestone
Hi,
I hope you don't mind me asking a novice question. I was an RSX-11M guru rather than a
VAX one - I never did much on the new fangled things.
However, when looking to set up a simh VAX the other day, I see that there are licences
required - how would I as a hobbyis tget one? Do I still have to try to joing DECUS (if it
still exists) or what is the process for requesting a hobbyist licence to bring up a VAX
node on HECnet?
Thanks
Peter
On 09/03/2023 03:39, Wilm Boerhout wrote:
I am running VAXeln on a simh node. It is just sitting
there purring and being proud of itself.
Along with the rest, I am enjoying HECnet so much. Thank you Johnny and all you other
enthusiasts.
Groet,
Wilm
(Verstuurd vanaf mijn telefoon, dus wat korter dan gewoonlijk.) (Sent
from my phone, so a bit more compact than usual)
_______________________________________________
HECnet mailing list -- hecnet(a)lists.dfupdate.se To unsubscribe send an email to
hecnet-leave(a)lists.dfupdate.se
Op 9 mrt. 2023 om 01:48 heeft Robert Armstrong
<bob(a)jfcl.com> het volgende geschreven:
_______________________________________________
HECnet mailing list -- hecnet(a)lists.dfupdate.se To unsubscribe send an
email to hecnet-leave(a)lists.dfupdate.se
DTR> sum 1 by os
Who has VAX Elan
running? What's that machine doing?
I think we should drop Cisco IOS and pyDECnet from the count -
those aren't really OSes for our purposes.
Bob
_______________________________________________
HECnet mailing list -- hecnet(a)lists.dfupdate.se To unsubscribe send
an email to hecnet-leave(a)lists.dfupdate.se
11:27 a.m.
Well. It's more that VSI never had a hobbyist license for VAX. That was
all HP, and when HP stopped their VMS hobbyist program, VAX lost.
There are several options, but no perfect ones.
1. Find an existing VMS license for VAX and use that. This is probably
the only legal option available.
2. Use an old enough version of VMS that don't have license checking
built in.
3. Patch VMS to circumvent the license checking.
4. Generate your own license, assuming you have the tools for it.
I don't encourage doing things that might be illegal. I also do not know
exactly what the legal status of different options are, and I am not a
lawyer.
But that's basically the situation for VAX people with VMS now.
Johnny
On 2023-03-09 11:19, Wilm Boerhout wrote:
Peter,
The VAX Hobbyist license program has been discontinued by the new owners (VSI) since
somewhere last year.
The only practical (legal?) option is now to run VMS V4.6, that precedes the PAK license
program.
I am not going to tell you about the VMS Liberation Front, because thatwould be against
the law.
Wilm
-----Original Message-----
From: Peter Whisker <peter(a)whisker.org.uk>
Sent: Thursday, March 9, 2023 11:16 AM
To: hecnet(a)lists.dfupdate.se
Subject: [HECnet] Re: A milestone
Hi,
I hope you don't mind me asking a novice question. I was an RSX-11M guru rather than
a VAX one - I never did much on the new fangled things.
However, when looking to set up a simh VAX the other day, I see that there are licences
required - how would I as a hobbyis tget one? Do I stillhave to try to joing DECUS (if it
still exists) or what is the process for requesting a hobbyist licence to bring up a VAX
node on HECnet?
Thanks
Peter
On 09/03/2023 03:39, Wilm Boerhout wrote:
--
Johnny Billquist || "I'm on a bus
|| on a psychedelic trip
email: bqt(a)softjar.se || Reading murder books
pdp is alive! || tryin' to stay hip" - B. Idol
I am running VAXeln on a simh node. It is just
sitting there purring and being proud of itself.
Along with the rest, I am enjoying HECnet so much. Thank you Johnny and all you other
enthusiasts.
Groet,
Wilm
(Verstuurd vanaf mijn telefoon, dus wat korter dan gewoonlijk.) (Sent
from my phone, so a bit more compact than usual)
_______________________________________________
HECnet mailing list -- hecnet(a)lists.dfupdate.se To unsubscribe send an email to
hecnet-leave(a)lists.dfupdate.se
_______________________________________________
HECnet mailing list -- hecnet(a)lists.dfupdate.se
To unsubscribe send an email to hecnet-leave(a)lists.dfupdate.se Op 9 mrt. 2023 om 01:48 heeft Robert Armstrong
<bob(a)jfcl.com> het volgende geschreven:
_______________________________________________
HECnet mailing list -- hecnet(a)lists.dfupdate.se To unsubscribe send an
email to hecnet-leave(a)lists.dfupdate.se DTR> sum 1 by os
Who has VAX Elan
running? What's that machine doing?
I think we should drop Cisco IOS and pyDECnet from the count -
those aren't really OSes for our purposes.
Bob
_______________________________________________
HECnet mailing list -- hecnet(a)lists.dfupdate.se To unsubscribe send
an email to hecnet-leave(a)lists.dfupdate.se 
12:47 p.m.
-----Original Message-----
From: Johnny Billquist <bqt(a)softjar.se>
Sent: Thursday, March 9, 2023 10:27 AM
To: hecnet(a)lists.dfupdate.se
Subject: [HECnet] Re: A milestone
Well. It's more that VSI never had a hobbyist license for VAX. That was all HP, and
when HP stopped their VMS hobbyist program, VAX lost.
There are several options, but no perfect ones.
1. Find an existing VMS license for VAX and use that. This is probably the only
legal option available.
But only if you pay HP to transfer it. HP will also sell you a re-cycled commercial
licence but presumably far to expensive for hobbyist use.
2. Use an old enough version of VMS that don't
have license checking built in.
3. Patch VMS to circumvent the license checking.
4. Generate your own license, assuming you have the tools for it.
5. Use a licence someone else generated.
6. If you have an expired hobbyist licence set the system date to be within the licence
period.
I don't encourage doing things that might be
illegal. I also do not know exactly
what the legal status of different options are, and I am not a lawyer.
The legality may also vary from country to country...
But that's basically the situation for VAX people
with VMS now.
Johnny
Dave
On 2023-03-09 11:19, Wilm Boerhout wrote:
_______________________________________________
HECnet mailing list -- hecnet(a)lists.dfupdate.se To unsubscribe send an
email to hecnet-leave(a)lists.dfupdate.se
_______________________________________________
HECnet mailing list -- hecnet(a)lists.dfupdate.se To unsubscribe send an
email to hecnet-leave(a)lists.dfupdate.se
--
Johnny Billquist || "I'm on a bus
|| on a psychedelic trip
email: bqt(a)softjar.se || Reading murder books
pdp is alive! || tryin' to stay hip" - B. Idol
_______________________________________________
HECnet mailing list -- hecnet(a)lists.dfupdate.se To unsubscribe send an email to
hecnet-leave(a)lists.dfupdate.se
Peter,
The VAX Hobbyist license program has been discontinued by the new owners
(VSI)
since somewhere last year.
The only practical (legal?) option is now to run VMS V4.6, that precedes the PAK
license program.
I am not going to tell you about the VMS Liberation Front, because thatwould be
against the law.
Wilm
-----Original Message-----
From: Peter Whisker <peter(a)whisker.org.uk>
Sent: Thursday, March 9, 2023 11:16 AM
To: hecnet(a)lists.dfupdate.se
Subject: [HECnet] Re: A milestone
Hi,
I hope you don't mind me asking a novice question. I was an RSX-11M guru
rather than a VAX one - I never did much on the new fangled things.
However, when looking to set up a simh VAX the other day, I see that there are
licences required - how would I as a hobbyis tget one? Do I stillhave to try to joing
DECUS (if it still exists) or what is the process for requesting a hobbyist licence to
bring up a VAX node on HECnet?
Thanks
Peter
On 09/03/2023 03:39, Wilm Boerhout wrote:
> I am running VAXeln on a simh node. It is just sitting there purring and being
proud of itself.
>
> Along with the rest, I am enjoying HECnet so much. Thank you Johnny and all
you other enthusiasts.
>
> Groet,
> Wilm
>
> (Verstuurd vanaf mijn telefoon, dus wat korter dan gewoonlijk.) (Sent
> from my phone, so a bit more compact than usual)
>
>> Op 9 mrt. 2023 om 01:48 heeft Robert Armstrong <bob(a)jfcl.com> het
volgende geschreven:
> DTR> sum 1 by os
Who has VAX Elan running? What's that machine doing?
I think we should drop Cisco IOS and pyDECnet from the count -
those aren't really OSes for our purposes.
Bob
_______________________________________________
HECnet mailing list -- hecnet(a)lists.dfupdate.se To unsubscribe send
an email to hecnet-leave(a)lists.dfupdate.se
_______________________________________________
HECnet mailing list -- hecnet(a)lists.dfupdate.se To unsubscribe send
an email to hecnet-leave(a)lists.dfupdate.se
5:04 p.m.
Johnny Billquist <bqt(a)softjar.se> wrote:
1. Find an existing VMS license for VAX and use that.
If you have real hardware then it probably (though not for sure) had a VMS license at
one time. So if you acquire an old VAX somehow, it's important to try to preserve the
existing contents of the system disk. If you can recover the LMF$LICENSE.LDB file then
you can reuse that with a new VMS installation.
Also, if it's a machine from the late 80s then it probably had actual paper PAKs
("Product Authorization Keys"). These were printed on single sheets of fancy
paper in neat little folders. If you acquire an old machine it's worth
checking/asking for those.
As for the legality of this, PAKs are transferrable with the machines, however getting
HP to do this for you today is likely impossible.
Bob
5:16 p.m.
Hi Bob,
I have written a few small programs in EPascal, such a DECnet client program to get the
time from a VMS system and set the ELN time from that. Also a LAT program that runs the
Edisplay program in a dedicated LAT service.
As for the kits, they were in some remote archive. I can send them to you. As for the
licenses, they were hobbyist so I no longer have them.
You are welcome to have a look. Set host to PIRTVX, any username will do.
Groet,
Wilm
(Verstuurd vanaf mijn telefoon, dus wat korter dan gewoonlijk.)
(Sent from my phone, so a bit more compact than usual)
Op 9 mrt. 2023 om 17:04 heeft Robert Armstrong
<bob(a)jfcl.com> het volgende geschreven:
Johnny Billquist <bqt(a)softjar.se> wrote:
1. Find an existing VMS license for VAX and use that.
If you have real hardware then it probably (though not for sure) had a VMS license at
one time. So if you acquire an old VAX somehow, it's important to try to preserve the
existing contents of the system disk. If you can recover the LMF$LICENSE.LDB file then
you can reuse that with a new VMS installation.
Also, if it's a machine from the late 80s then it probably had actual paper PAKs
("Product Authorization Keys"). These were printed on single sheets of fancy
paper in neat little folders. If you acquire an old machine it's worth
checking/asking for those.
As for the legality of this, PAKs are transferrable with the machines, however getting
HP to do this for you today is likely impossible.
Bob
_______________________________________________
HECnet mailing list -- hecnet(a)lists.dfupdate.se
To unsubscribe send an email to hecnet-leave(a)lists.dfupdate.se
14 Apr
14 Apr
4:59 p.m.
New subject: VSI VMS x86 on HECnet
Only a little work was needed to put another VSI x86 node on HECnet.
NLx86 (29.213, registry pending) is a guest of VirtualBox 7.0.6 on my HP
Zbook 15u G4 laptop (from 2017)
It's surprisingly fast, considering the virtualization in between, and the
age of the laptop.
Wilm
5:13 p.m.
New subject: VSI VMS x86 on HECnet
Wilm,
On any fairly recent CPU you should get near native performance when running
a Virtual Machine. Both AMD and INTEL CPUs have instructions to support
virtualization so there is no need to emulate instructions in the way SIMH
or the Alpha Emulators do.
Dave
-----Original Message-----
From: Wilm Boerhout <wboerhout(a)gmail.com>
Sent: Friday, April 14, 2023 4:00 PM
To: 'The Hobbyist DECnet mailing list' <hecnet(a)lists.dfupdate.se>
Subject: [HECnet] VSI VMS x86 on HECnet
Only a little work was needed to put another VSI x86 node on HECnet.
NLx86 (29.213, registry pending) is a guest of VirtualBox 7.0.6 on my HP
Zbook
15u G4 laptop (from 2017)
It's surprisingly fast, considering the virtualization in between, and the
age
of the
laptop.
Wilm
_______________________________________________
HECnet mailing list -- hecnet(a)lists.dfupdate.se To unsubscribe send an
email to
hecnet-leave(a)lists.dfupdate.se
7:01 p.m.
New subject: VSI VMS x86 on HECnet
What instructions did you have in mind?
The reason I'm asking is because virtualization is a very old concept
(in computer terms) dating to the late 1960's. You don't actually need
special instructions, just to handle a trap that would take you out of
problem state (like doing I/O). I /think/// you need the VM bit on x86
to handle real mode applications and operating systems (I.E., DOS)
because you don't get the trap. Was there something else?
IBM enhanced their hardware to provide more support for virtualization
and these are called LPAR's.
The unreleased Jupiter instantiation of the PDP-10 architecture also had
a VM bit; again, I /think/ to support user mode applications which
expected unrestricted access to hardware. Tops-10 supports this and
Tops-20 has a limited form of it (USRIO%). In any event, it would have
allowed you to boot Tops-20 under Tops-20, a fine concept.
I don't recall that either the VAX or Alpha supported virtualization.
One imagines that perhaps this was thought unnecessary because the
available internal instruction set simulators and the lowering cost of
the hardware.
------------------------------------------------------------------------
On 4/14/23 11:13 AM,dave.g4ugm(a)gmail.com wrote:
Wilm,
On any fairly recent CPU you should get near native performance when running
a Virtual Machine. Both AMD and INTEL CPUs have instructions to support
virtualization so there is no need to emulate instructions in the way SIMH
or the Alpha Emulators do.
Dave
> ------------------------------------------------------------------------
> *From*: Wilm Boerhout<wboerhout(a)gmail.com>
> *Sent*: Friday, April 14, 2023 4:00 PM
> *To*: 'The Hobbyist DECnet mailing list'<hecnet(a)lists.dfupdate.se>
> *Subject*: [HECnet] VSI VMS x86 on HECnet
>
> Only a little work was needed to put another VSI x86 node on HECnet.
>
> NLx86 (29.213, registry pending) is a guest of VirtualBox 7.0.6 on my HP Zbook15u G4
laptop (from 2017)
>
> It's surprisingly fast, considering the virtualization in between, and the
> age of the laptop.
>
> Wilm
8:49 p.m.
New subject: VSI VMS x86 on HECnet
On Apr 14, 2023, at 1:01 PM, Thomas DeBellis
<tommytimesharing(a)gmail.com> wrote:
...
I don't recall that either the VAX or Alpha supported virtualization. One imagines
that perhaps this was thought unnecessary because the available internal instruction set
simulators and the lowering cost of the hardware.
I don't see virtualization in the Alpha architecture manual, but it can be found in
chapter 12 of DEC Std 032, the VAX standard.
paul
9:03 p.m.
New subject: VSI VMS x86 on HECnet
Thomas,
Lets start with the stuff every one else wants to read. The overhead of trap handling ,
especially where you are running two level virtual memory, is significant.
So on S/390 IBM introduced the SIE instruction (Start Interpretive Execution). The
equivalent Intel and AMD features are called VT-X and AMD-V respectively.
These allow the CPU microcode to virtually execute some privileged instructions that would
otherwise require a trap and context switch so slowing things down.
You will find that VirtualBox requires these extensions, and it won’t start if they are
not enabled..
Not every architecture can be virtualized by trapping as you describe. The requirements
are defined by the Popek and Goldberg requirements.
https://en.wikipedia.org/wiki/Popek_and_Goldberg_virtualization_requirements
So S/370 and Motorola 68010 are a couple of CPUs that satisfy them. The M68000 and the
370XA, ESA390, IBM z don’t….
.. I believe the same applies to X86/X64 so if you want to virtualize these without VT-X
or AMD-V you need to build a SIMH style emulator.
Early versions of VMWare Workstation worked this way, and it was slow. I know I used it
:-)
A modern tool which does this is DOSBox which is why it works on other platforms
Dave
From: Thomas DeBellis <tommytimesharing(a)gmail.com>
Sent: Friday, April 14, 2023 6:02 PM
To: hecnet(a)lists.dfupdate.se
Subject: [HECnet] Re: VSI VMS x86 on HECnet
What instructions did you have in mind?
The reason I'm asking is because virtualization is a very old concept (in computer
terms) dating to the late 1960's. You don't actually need special instructions,
just to handle a trap that would take you out of problem state (like doing I/O). I think
you need the VM bit on x86 to handle real mode applications and operating systems (I.E.,
DOS) because you don't get the trap. Was there something else?
IBM enhanced their hardware to provide more support for virtualization and these are
called LPAR's.
The unreleased Jupiter instantiation of the PDP-10 architecture also had a VM bit; again,
I think to support user mode applications which expected unrestricted access to hardware.
Tops-10 supports this and Tops-20 has a limited form of it (USRIO%). In any event, it
would have allowed you to boot Tops-20 under Tops-20, a fine concept.
I don't recall that either the VAX or Alpha supported virtualization. One imagines
that perhaps this was thought unnecessary because the available internal instruction set
simulators and the lowering cost of the hardware.
_____
On 4/14/23 11:13 AM, dave.g4ugm(a)gmail.com <mailto:dave.g4ugm@gmail.com> wrote:
Wilm,
On any fairly recent CPU you should get near native performance when running
a Virtual Machine. Both AMD and INTEL CPUs have instructions to support
virtualization so there is no need to emulate instructions in the way SIMH
or the Alpha Emulators do.
Dave
_____
From: Wilm Boerhout <mailto:wboerhout@gmail.com> <wboerhout(a)gmail.com>
Sent: Friday, April 14, 2023 4:00 PM
To: 'The Hobbyist DECnet mailing list' <mailto:hecnet@lists.dfupdate.se>
<hecnet(a)lists.dfupdate.se>
Subject: [HECnet] VSI VMS x86 on HECnet
Only a little work was needed to put another VSI x86 node on HECnet.
NLx86 (29.213, registry pending) is a guest of VirtualBox 7.0.6 on my HP Zbook15u G4
laptop (from 2017)
It's surprisingly fast, considering the virtualization in between, and the
age of the laptop.
Wilm
5:34 p.m.
New subject: VSI VMS x86 on HECnet
Given my current config, I’ve been quite pleased with the performance, especially as I’m
running on a 9 year old CPU (2014). Before long I hope to upgrade to a DL380 Gen9.
Hypervisor: VMware ESXi, 6.7.0
Model: HP ProDesk 600 G1 SFF
Processor Type: Intel(R) Core(TM) i5-4590 CPU @ 3.30GHz
Logical Processors: 4
Zane
On Apr 14, 2023, at 7:59 AM, Wilm Boerhout
<wboerhout(a)gmail.com> wrote:
Only a little work was needed to put another VSI x86 node on HECnet.
NLx86 (29.213, registry pending) is a guest of VirtualBox 7.0.6 on my HP
Zbook 15u G4 laptop (from 2017)
It's surprisingly fast, considering the virtualization in between, and the
age of the laptop.
Wilm
_______________________________________________
HECnet mailing list -- hecnet(a)lists.dfupdate.se
To unsubscribe send an email to hecnet-leave(a)lists.dfupdate.se
5:37 p.m.
New subject: VSI VMS x86 on HECnet
I've built my second and like you surprised/pleased by the performance
on relatively modest hardware. KVM and their ilk only put a tiny amount
of overhead into the mix these days. However, I was secretly expecting
to need a full 8+ cores and 64GB RAM.....but nope.
I'm running on KVM under ubuntu (basically debian+KVM), I can't get it
running under KVM on Proxmox which is a bit surprising (basically
debian+KVM). I might revisit that.
M
On 14/04/2023 15:59, Wilm Boerhout wrote:
Only a little work was needed to put another VSI x86
node on HECnet.
NLx86 (29.213, registry pending) is a guest of VirtualBox 7.0.6 on my HP
Zbook 15u G4 laptop (from 2017)
It's surprisingly fast, considering the virtualization in between, and the
age of the laptop.
Wilm
_______________________________________________
HECnet mailing list -- hecnet(a)lists.dfupdate.se
To unsubscribe send an email to hecnet-leave(a)lists.dfupdate.se
9 Mar
9 Mar
4:46 p.m.
I can only speak for 60.*
CAMECA - PDP-11/73 (needs work to be brought back online)
ZEUS - Linux PC, no clue if it will even boot - pretty sure it’s never been on HECnet
SAVAGE - VM - currently offline, not really used, and didn’t get brought up after last
power outage
JOHNNY - VAXstation 4000/90 - offline - I had it online during my last Sabbatical in 2021,
but finally put it away as I needed the space
MONK - VM
HAM - VM
DOC - I think this is my MicroVAX III (that I don’t know says a lot, not sure what it
would take to bring it up) - pretty sure it’s never been on HECnet
RENNY - VM
LNGTOM - VM
DS201 - DECserver - offline - pretty sure it’s never been on HECnet and I’m not even sure
why it was assigned an address, at least I know where it is. :-)
RSTSE - PDP-11/73 (needs work to be brought back online)
PDXVAX - VAXstation 4000/60 - DECnet Router, it’s online, unless it’s too hot to run it
SAVAGE, JOHNNY, MONK, HAM, RENNY, and LNGTOM are all part of the same cluster. Getting
them virtualized did wonders for my power bill. Especially MONK, which had a lot of
external disks and tape drives.
Zane
On Mar 8, 2023, at 3:41 PM, Robert Armstrong
<bob(a)jfcl.com> wrote:
I saw 999 nodes this morning when I updated my database - somebody else
must have signed up today :)
Pretty neat, but as you say I wonder how many names are just reserved for
future expansion, and how many really exist?
And I wonder how many are real, physical, machines and how many are
simulations?
And how many are real machines, but not actual DEC machines? (yes, there
are some!)
It'd be very interesting to see some statistics...
Bob
_______________________________________________
HECnet mailing list -- hecnet(a)lists.dfupdate.se
To unsubscribe send an email to hecnet-leave(a)lists.dfupdate.se
4:34 p.m.
Congratulations! I know some of us started out on HECnet a long time ago, I don’t
remember when I joined, only that it was well prior to 2008. I also know that I was
mostly offline from late 2008 until something like 2016. I have more systems online now
than I did prior to 2008. Though one of my nodes has only been online a few times.
Zane
On Mar 8, 2023, at 12:47 PM, Johnny Billquist
<bqt(a)softjar.se> wrote:
I have been planning for a long time to post a small reflection when I reached 1000 nodes
registered in the hecnet nodename database.
10 Mar
10 Mar
6:30 a.m.
Thank you, Johnny! An amazing network and a great community you have
created. Well done!
Regards, Tim.
On Thu, 9 Mar 2023 at 04:47, Johnny Billquist <bqt(a)softjar.se> wrote:
I have been planning for a long time to post a small
reflection when I
reached 1000 nodes registered in the hecnet nodename database.
This happened tonight. I think it is a pretty cool thing. There are now
1000 nodenames registered in this small hobby DECnet. I guess you could
say it's actually not that small.
However, I know that rather few machines are actually online, and it
might even be that the majority of nodes registered have never been
online. My guess is that maybe 5% of the registered machines are usually
online. But I do not have any hard data to back this up.
HECnet started out about 20 years ago from a desire I had to hook up a
PDP-11 I had at home, to some machines at my university, using DECnet.
At the time, I didn't have any TCP/IP for RSX, and the only way to get
any kind of networking was to try and come up with some way of getting
DECnet up.
My first implementation basically just forwarded a serial port
communication between two Unix machines. And on each end I then hooked
that serial port into a PDP-11 running RSX, and used DDCMP for the
actual DECnet link.
This worked, but was obviously not that fast, as the serial ports were
limited to 9600 bps.
After a year or so, I figured I could instead write a small program that
would forward ethernet packets. Using UDP I basically had the same
property as a local ethernet, but it could be located somewhere pretty
far away. From a DECnet point of view, it would appear as if they were
on the same ethernet segment, while in reality they were nowhere near.
As DECnet have pretty long timeouts on things, it turned out this worked
without a hitch, and I could achieve much better throughput.
Not long after that, the first other users were hooked up to HECnet as
well. This was maybe around 2003 or so.
Another data point is that there are 113 different persons that have one
or more nodenames registered.
Happy milestone, everyone, and thanks for being around.
Johnny
--
Johnny Billquist || "I'm on a bus
|| on a psychedelic trip
email: bqt(a)softjar.se || Reading murder books
pdp is alive! || tryin' to stay hip" - B. Idol
_______________________________________________
HECnet mailing list -- hecnet(a)lists.dfupdate.se
To unsubscribe send an email to hecnet-leave(a)lists.dfupdate.se
380
days inactive
417
days old
74 comments
21 participants
participants (21)
-
Brian Angus -
Brian Hechinger -
Charles Guldenschuh -
Dave McGuire -
dave.g4ugm＠gmail.com -
Ian McLaughlin -
John Yaldwyn -
Johnny Billquist -
Keith Halewood -
Mark Curtis -
Mark J. Blair -
Mark Matlock -
Mike Kostersitz -
Paul Koning -
Peter Whisker -
Robert Armstrong -
Supratim Sanyal -
Thomas DeBellis -
Tim Sneddon -
Wilm Boerhout -
Zane Healy