22 Oct
2020
22 Oct
'20
10:30 p.m.
Christmas Eve, 2019 :)
On Dec 24, 2019, at 8:56 PM, Paul Koning
<paulkoning at comcast.net> wrote:
I?ve never seen a Phase I protocol spec but DECnet/8 was probably closer to Phase
I than Phase II although the dates in the source code appear to cover both phases. Even
after Phase II development started the protocols were still being updated - datagram
service was still in the NSP spec in early 1977 but was dropped about half way through the
development cycle. One advantage we had for DECnet-RSX development (11M/11S/11D and IAS)
was that they all shared common protocol processing code, which had been written for an
earlier advanced development project, so protocol changes were relatively easy to make.
John.
---
Supratim Sanyal, W1XMT
39.19151 N, 77.23432 W
QCOCAL::SANYAL via HECnet
On Dec 24, 2019, at 11:59 AM, John Forecast
<john at forecast.name> wrote:
A while ago I found a Phase I RTS-8 document that describes the protocol in reasonable
detail. It's clear that Phase I has only a vague resemblance to the later protocols.
For example, the NSP protocol is there, but it's seriously different. For example, it
offers both a connection service and a datagram service. And various other things work
entirely differently than the later versions. So while Phase III and IV can talk to Phase
II NSP with no real trouble, there is no way for any of them to make sense out of what a
Phase I node is saying.
If you have source code, it should be pretty obvious if it's Phase I or Phase II from
looking at the NSP packet formats.
On Dec 23, 2019, at 5:51 PM, Johnny Billquist
<bqt at softjar.se> wrote:
Yes, I agree. I just took a look at the code and the early dates match when Phase
I was available (I was using it around the middle of 1976) and the later dates (mid-1977)
would have been when we were finalizing the DECnet-RSX Phase II system architecture. The
code claims to support NSP version 2.2 but I have no way to map that to a particular
phase, only that version 1.0 was dated July 1975 and version 3.1 (Phase III) was dated
March 1978. On 2019-12-23 20:24, John Forecast wrote:
> ...
That was implemented on RTS-8 and looks like a Phase I implementation - all hand-crafted
PAL code. The floppies are available on the net and includes full source code. When I
joined the DECnet development group in early 1977, there were a couple of PDP-8 developers
as part of the group. I don?t know if they were developing a Phase II implementation but
they disappeared after about 6 months - not surprising given the difficulties we were
having getting it to fit in a 28KW PDP-11.
Poul took a look at the DECNET-8 sources, and figured it was actually not phase I (I
thought it was phase I as well). Poul thinks it's close to, if not actually phase II.
There are things in there that apparently did not exist in phase I.
On Oct 22, 2020, at 4:14 PM, Paul Koning
<paulkoning at comcast.net> wrote:
On Oct 22, 2020, at 3:43 PM, R. Voorhorst
<R.Voorhorst at swabhawat.com> wrote:
@John Forecast:
Hi John,
The oddity is just the other way around ?.
The release date of the primitive software, say Phase-I+ is dated at 8-april-1977.
The Nip printout is dated 28-dec-1976 which aligns with your story.
So why ship something clearly broken with an obvious simple bug that should have been
easily catched with a minor level of release quality control.
When the Pdp8 development was dropped around may-1977, why not release the current stuff
instead?
Some form of date editing in the software was clearly done on package level as the date
8-apr-77 appears all over the place, but why for a product at that time already 1-3 years
ago?
My memory must be fading, because the Phase I vs. Phase II discussion Johnny referred to
doesn't ring bells.
Given that you have actual software and it's approaching running status, we'll
soon know by examination. As soon as you start making NSP connections, it's very
clear. This is where Phase I and Phase II are different and incompatible.
paul