19 Nov
2021
19 Nov
'21
12:35 a.m.
No, the Dup is only simulated as Kdp/Dup combo, hence the Ddcmp characteristics of the
pair.
Reindert
-----Original Message-----
From: owner-hecnet at Update.UU.SE [mailto:owner-hecnet at Update.UU.SE] On Behalf Of
Johnny Billquist
Sent: Friday, 19 November, 2021 00:31
To: hecnet at Update.UU.SE
Subject: Re: [HECnet] ANF10 network --> DN200 (RJE, Decnet) --> should be Ddcmp
--Re2
Fair enough.
But DUP11 is a sync character based interface. So that one too requires that you do DDCMP
in software. But I didn't think simh simulated a DUP11?
Johnny
On 2021-11-18 23:33, R. Voorhorst wrote:
Yes, but they are async lines and these are indeed
character based with software determining the protocol, not so the sync liness. Async
decnet ddcmp behaves slightly differently from sync decnet ddcmp in number of characters
used. Look at the packet contents in used characters and in between.
Reindert
-----Original Message-----
From: owner-hecnet at Update.UU.SE [mailto:owner-hecnet at Update.UU.SE] On
Behalf Of Johnny Billquist
Sent: Thursday, 18 November, 2021 23:23
To: hecnet at Update.UU.SE
Subject: Re: [HECnet] ANF10 network --> DN200 (RJE, Decnet) --> should
be Ddcmp --Re
Well... RSX for example can talk DDCMP over DZ11 or DL11 lines just as well, which are
plain simple serial lines existing in simh.
Hopefully such lines can be connected with interfaces doing DDCMP in hardware (simulated
in simh of course), and it should all work.
And any software running in RSX just have an interface that gives the DDCMP packet layer.
The actual serial lines are not accessible as such by any other software than the DDCMP
driver.
Johnny
On 2021-11-18 23:01, R. Voorhorst wrote:
--
Johnny Billquist || "I'm on a bus
|| on a psychedelic trip
email: bqt at softjar.se || Reading murder books
pdp is alive! || tryin' to stay hip" - B. Idol
Point in this case is that the ddcmp is handled
in Simh and not in system software so software can only deal with packets/messages; there
is need for a simh sync line where the software itself will be able to communicate in
bytes/characters and synthesize the packets/messages as the next higher organization.
Reindert
-----Original Message-----
From: owner-hecnet at Update.UU.SE [mailto:owner-hecnet at Update.UU.SE] On
Behalf Of Johnny Billquist
Sent: Thursday, 18 November, 2021 22:39
To: hecnet at Update.UU.SE
Subject: Re: [HECnet] ANF10 network --> DN200 (RJE, Decnet) -->
should be Ddcmp
Well, a serial line over which DDCMP is running is purely a packet based interface.
Johnny
On 2021-11-18 21:46, R. Voorhorst wrote:
Indeed it is Ddcmp, but the meaning is the
difference between
packet/message oriented and character oriented and the latter should also exist.
Reindert
-----Original Message-----
From: owner-hecnet at Update.UU.SE [mailto:owner-hecnet at Update.UU.SE]
On Behalf Of Robert Armstrong
Sent: Thursday, 18 November, 2021 21:11
To: hecnet at Update.UU.SE
Subject: RE: [HECnet] ANF10 network --> DN200 (RJE, Decnet)
Paul Koning <paulkoning at comcast.net>
wrote:
They speak DDCMP, yes, but not DECnet. Any protocol layered over
DDCMP
should work.
Indeed, and we just proved that with ANF10 which uses the DMCs
and DUPs but is not DECnet.
Bob
19 Nov
19 Nov
1:12 a.m.
It seems like a very simple matter to simulate it as a basic character device, separate
from the KMC pairing. The only real question is how you represent a sync data stream in a
simullated device. Is it just the byte stream over a TCP connection? That would be easy
for synchronous character mode. If you also want to simulate HDLC sync devices,
that's a different matter entirely; there isn't an obvious way to do that.
In other words, if someone wants this, just design how it communicates and do the code, it
doesn't sound hard.
paul
On Nov 18, 2021, at 6:35 PM, R. Voorhorst
<R.Voorhorst at swabhawat.com> wrote:
No, the Dup is only simulated as Kdp/Dup combo, hence the Ddcmp characteristics of the
pair.
Reindert
-----Original Message-----
From: owner-hecnet at Update.UU.SE [mailto:owner-hecnet at Update.UU.SE] On Behalf Of
Johnny Billquist
Sent: Friday, 19 November, 2021 00:31
To: hecnet at Update.UU.SE
Subject: Re: [HECnet] ANF10 network --> DN200 (RJE, Decnet) --> should be Ddcmp
--Re2
Fair enough.
But DUP11 is a sync character based interface. So that one too requires that you do DDCMP
in software. But I didn't think simh simulated a DUP11?
Johnny
1:50 a.m.
Correction...
Character mode sync in SIMH seems utterly trivial; just treat it as an async stream except
that the device would supply sync bytes when the software isn't feeding it data for a
while. Perhaps just a couple, just enough to act as a boundary; no need to flood the
connection with data.
HDLC mode is different; HDLC is inescapably a packet mode service since the flags are
packet delimiters. A small complication is that it can carry any bit count, not just
whole bytes. So the network encoding of a simulated HDLC device needs to handle that. A
UDP encoding could just be a packet containing the data bytes, with one byte appended to
specify how many bits in the last data byte are meaningful. (So a typical whole-byte
packet might be "abcdefgh\010".) A TCP encoding could be similar with a packet
length at the start. If it's desirable not to buffer whole frames, the stream could
be broken into fragments with the last one (where the flag appears) marked as last. That
would be necessary if arbitrary length frames are wanted, though a sensible protocol
doesn't use frames longer than a thousand bytes or so since the 16-bit CRC isn't
strong enough for really long frames.
paul
On Nov 18, 2021, at 7:12 PM, Paul Koning
<paulkoning at comcast.net> wrote:
It seems like a very simple matter to simulate it as a basic character device, separate
from the KMC pairing. The only real question is how you represent a sync data stream in a
simullated device. Is it just the byte stream over a TCP connection? That would be easy
for synchronous character mode. If you also want to simulate HDLC sync devices,
that's a different matter entirely; there isn't an obvious way to do that.
In other words, if someone wants this, just design how it communicates and do the code,
it doesn't sound hard.
paul
> On Nov 18, 2021, at 6:35 PM, R. Voorhorst <R.Voorhorst at swabhawat.com>
wrote:
>
> No, the Dup is only simulated as Kdp/Dup combo, hence the Ddcmp characteristics of
the pair.
>
>
> Reindert
>
> -----Original Message-----
> From: owner-hecnet at Update.UU.SE [mailto:owner-hecnet at Update.UU.SE] On Behalf Of
Johnny Billquist
> Sent: Friday, 19 November, 2021 00:31
> To: hecnet at Update.UU.SE
> Subject: Re: [HECnet] ANF10 network --> DN200 (RJE, Decnet) --> should be Ddcmp
--Re2
>
> Fair enough.
>
> But DUP11 is a sync character based interface. So that one too requires that you do
DDCMP in software. But I didn't think simh simulated a DUP11?
>
> Johnny
11:58 a.m.
Folks,
It might be worth looking at how Hercules encapsulates SDLC as that will do
sync mode.
There are also RFCs for these processes e.g. RFC4349.
Could we use any of these?
Dave
-----Original Message-----
From: owner-hecnet at Update.UU.SE <owner-hecnet at Update.UU.SE> On
Behalf Of Paul Koning
Sent: 19 November 2021 00:51
To: hecnet at Update.UU.SE
Subject: Re: [HECnet] ANF10 network --> DN200 (RJE, Decnet) --> should be
Ddcmp --Re3
Correction...
Character mode sync in SIMH seems utterly trivial; just treat it as an
async
stream except that the device would supply sync bytes
when the software
isn't feeding it data for a while. Perhaps just a couple, just enough to
act
as a
boundary; no need to flood the connection with data.
HDLC mode is different; HDLC is inescapably a packet mode service since
the
flags are packet delimiters. A small complication is
that it can carry
any bit
count, not just whole bytes. So the network encoding
of a simulated HDLC
device needs to handle that. A UDP encoding could just be a packet
containing the data bytes, with one byte appended to specify how many bits
in the last data byte are meaningful. (So a typical whole-byte packet
might
be "abcdefgh\010".) A TCP encoding could
be similar with a packet length
at
the start. If it's desirable not to buffer whole
frames, the stream could
be
broken into fragments with the last one (where the
flag appears) marked as
last. That would be necessary if arbitrary length frames are wanted,
though a
sensible protocol doesn't use frames longer than a
thousand bytes or so
since the 16-bit CRC isn't strong enough for really long frames.
paul
device,
On Nov 18, 2021, at 7:12 PM, Paul Koning
<paulkoning at comcast.net>
wrote:
>
> It seems like a very simple matter to simulate it as a basic character
separate from the KMC pairing. The only real question
is how you
represent
a sync data stream in a simullated device. Is it just
the byte stream
over a
TCP connection? That would be easy for synchronous
character mode. If
you also want to simulate HDLC sync devices, that's a different matter
entirely; there isn't an obvious way to do that.
In other words, if someone wants this, just design how it communicates
and do the
code, it doesn't sound hard.
paul
> On Nov 18, 2021, at 6:35 PM, R. Voorhorst
<R.Voorhorst at swabhawat.com>
wrote:
>
> No, the Dup is only simulated as Kdp/Dup combo, hence the Ddcmp
characteristics of the pair.
>
>
> Reindert
>
> -----Original Message-----
> From: owner-hecnet at Update.UU.SE [mailto:owner-
hecnet at Update.UU.SE] On
Behalf Of Johnny Billquist
> Sent: Friday, 19 November, 2021 00:31
> To: hecnet at Update.UU.SE
> Subject: Re: [HECnet] ANF10 network --> DN200 (RJE, Decnet) --> should
be Ddcmp --Re2
>
> Fair enough.
>
> But DUP11 is a sync character based interface. So that one too requires
that
you do DDCMP in software. But I didn't think simh simulated a DUP11?
>>
>> Johnny
4:14 p.m.
On Nov 19, 2021, at 5:58 AM, <dave.g4ugm at
gmail.com> <dave.g4ugm at gmail.com> wrote:
Folks,
It might be worth looking at how Hercules encapsulates SDLC as that will do
sync mode.
There are also RFCs for these processes e.g. RFC4349.
Could we use any of these?
Dave
That might make sense, though carrying along all the baggage of L2TP may be more trouble
than it's worth.
paul
5:32 p.m.
<dave.g4ugm at gmail.com> <dave.g4ugm at
gmail.com> wrote:
It might be worth looking at how Hercules encapsulates SDLC as that will
do
TOPS-10 had an IBM HASP communication FE product; I think it was called a
DN60. I used it long ago - you could submit card decks to JES2 under OS/VS2
from files on the -10 and get the resulting print out back as another disk
file on the -10. Beats a keypunch and a line printer by miles... It'd be
kind of neat, although a total waste of time, to hook up a simh TOPS10 to a
Hercules and try to get that working.
Bob
20 Nov
20 Nov
1:27 a.m.
On 11/19/21 11:32 AM, Robert Armstrong wrote:
The fact that it would be neat means, to me, that it's not a waste of
time.
Is that software around somewhere? I'd love to try this with real
iron at some point, at the museum.
-Dave
--
Dave McGuire, AK4HZ
New Kensington, PA
<dave.g4ugm
at gmail.com> <dave.g4ugm at gmail.com> wrote:
It might be worth looking at how Hercules encapsulates SDLC as that will
do
TOPS-10 had an IBM HASP communication FE product; I think it was called a
DN60. I used it long ago - you could submit card decks to JES2 under OS/VS2
from files on the -10 and get the resulting print out back as another disk
file on the -10. Beats a keypunch and a line printer by miles... It'd be
kind of neat, although a total waste of time, to hook up a simh TOPS10 to a
Hercules and try to get that working.
6:25 p.m.
On 11/19/21 7:52 PM, Robert Armstrong wrote:
Will do. I think it's a long shot, but you never know. We get boxes
of tapes in all the time.
-Dave
--
Dave McGuire, AK4HZ
New Kensington, PA
Is that
software around somewhere? I'd love to try this with real
iron at some point, at the museum.
I haven't seen it, but if you find a copy, let me know. 
19 Nov
19 Nov
5:18 a.m.
New subject: ANF10 network --> DN200 (RJE, Decnet) --> should be Ddcmp --Re3
I have not followed the details of this conversation, BUT I'm 99.99% sure that a PDP11
can be configured with just a DUP with DECnet using that interface, and that DUP can be
"wired" to another simulator which has a DUP, KDP/DUP or a DMC and DECnet can be
spoken just fine.
- Mark
On Thursday, November 18, 2021 at 3:35 PM, Reindert wrote:
No, the Dup is only simulated as Kdp/Dup combo, hence the Ddcmp
characteristics of the pair.
Reindert
-----Original Message-----
From: owner-hecnet at Update.UU.SE [mailto:owner-
hecnet at Update.UU.SE] On Behalf Of Johnny Billquist
Sent: Friday, 19 November, 2021 00:31
To: hecnet at Update.UU.SE
Subject: Re: [HECnet] ANF10 network --> DN200 (RJE, Decnet) --> should be
Ddcmp --Re2
Fair enough.
But DUP11 is a sync character based interface. So that one too requires that
you do DDCMP in software. But I didn't think simh simulated a DUP11?
Johnny
On 2021-11-18 23:33, R. Voorhorst wrote:
--
Johnny Billquist || "I'm on a bus
|| on a psychedelic trip
email: bqt at softjar.se || Reading murder books
pdp is alive! || tryin' to stay hip" - B. Idol
Yes, but they are async lines and these are
indeed character based with
software determining the protocol, not so the sync
liness. Async decnet
ddcmp behaves slightly differently from sync decnet ddcmp in number of
characters used. Look at the packet contents in used characters and in
between.
Reindert
-----Original Message-----
From: owner-hecnet at Update.UU.SE [mailto:owner-
hecnet at Update.UU.SE] On
Behalf Of Johnny Billquist
Sent: Thursday, 18 November, 2021 23:23
To: hecnet at Update.UU.SE
Subject: Re: [HECnet] ANF10 network --> DN200 (RJE, Decnet) --> should
be Ddcmp --Re
Well... RSX for example can talk DDCMP over DZ11 or DL11 lines just as well,
which
are plain simple serial lines existing in simh.
Hopefully such lines can be connected with
interfaces doing DDCMP in
hardware (simulated in simh of course), and it should all
work.
And any software running in RSX just have an interface that gives the
DDCMP packet
layer. The actual serial lines are not accessible as such by any
other software than the DDCMP driver.
Johnny
On 2021-11-18 23:01, R. Voorhorst wrote:
> Point in this case is that the ddcmp is handled in Simh and not in system
software so software can only deal with packets/messages; there is need for
a simh sync line where the software itself will be able to communicate in
bytes/characters and synthesize the packets/messages as the next higher
organization.
>
> Reindert
>
> -----Original Message-----
> From: owner-hecnet at Update.UU.SE [mailto:owner-
hecnet at Update.UU.SE] On
> Behalf Of Johnny Billquist
> Sent: Thursday, 18 November, 2021 22:39
> To: hecnet at Update.UU.SE
> Subject: Re: [HECnet] ANF10 network --> DN200 (RJE, Decnet) -->
> should be Ddcmp
>
> Well, a serial line over which DDCMP is running is purely a packet based
interface.
>
> Johnny
>
> On 2021-11-18 21:46, R. Voorhorst wrote:
>> Indeed it is Ddcmp, but the meaning is the difference between
>> packet/message oriented and character oriented and the latter should
also
exist.
>>
>> Reindert
>>
>> -----Original Message-----
>> From: owner-hecnet at Update.UU.SE [mailto:owner-
hecnet at Update.UU.SE]
On Behalf Of Robert Armstrong
Sent: Thursday, 18 November, 2021 21:11
To: hecnet at Update.UU.SE
Subject: RE: [HECnet] ANF10 network --> DN200 (RJE, Decnet)
> Paul Koning <paulkoning at comcast.net> wrote:
> They speak DDCMP, yes, but not DECnet. Any protocol layered over
> DDCMP
should work.
Indeed, and we just proved that with ANF10 which uses the DMCs
and DUPs but is not DECnet.
Bob
890
days inactive
892
days old
9 comments
6 participants
participants (6)
-
bob＠jfcl.com -
dave.g4ugm＠gmail.com -
Mark＠infocomm.com -
mcguire＠neurotica.com -
paulkoning＠comcast.net -
R.Voorhorst＠swabhawat.com