nls-restore
[Top] [All Lists]

Re: Restart of NLS/Augment effort

To: Philip Gust <gust@NouveauSystems.com>
Cc: Jonathan Cheyer <jonathan@cheyer.biz>, Ken Harrenstien <klh@panix.com>, "Peter P. Yim" <peter.yim@cim3.com>
From: Ken Harrenstien <klh@panix.com>
Date: Sun, 13 Feb 2005 10:17:59 -0800 (PST)
Message-id: <CMM.0.91.0.1108318672.klh@panix1.panix.com>
Commenting on various snippets:    (01)

> >I'd tend to agree with Phil that we should go with the two-step approach. 
> >It should be easier to start with a clone of Doug's system which would 
> >allow us to start using NLS and then we can learn how to build it from a 
> >clean TOPS20 as we learn the system.    (02)

A couple notes about vanilla/clean TOPS-20: DEC has made all of the
T10/T20 software available on a personal use basis.  However, there
have always been large amounts of non-DEC software and monitor
modifications floating around the early Arpanet/Internet which were
essential to any useful site on the network.  Mark Crispin has been
collecting this into what is intended to become a standard
distribution for T20 hackers, and I suggest starting with that once we
are ready for this step.    (03)

> >Ken, I understand that the easiest way to copy the whole system is 
> >just  to copy the disk images themselves, but of course that would also 
> >copy all Doug's personal files. Is there an easy way to copy specific sets 
> >of files off the TOPS20 file system? Is there a TOPS20 command that does 
> >something similar to a Linux find/tar combo piping the resulting output 
> >over a network connection to an external (i.e. Linux) file system?    (04)

There is a way, but it's not as simple as you might like.  You'd have
to run DUMPER and give it a selected list of directories; you can then
take the resulting virtual tape file and "mount" it on another system
for restoration.  The pain comes from having to decide which dirs to
save.  There are enough of them that what I suggest is to start with
full disk images and delete what you don't want to keep -- the
clone/prune approach.  Then, to increase portability and eliminate the
problem of disk blocks that were part of deleted files but still
present in the image, I suggest running DUMPER with a full wild-card
specification (<*>*.*.*) for each structure.  The resulting "tape"
will be clean and the contents can then be restored to different
structures, or into a Unix filesystem for easier perusal.    (05)

> >Our primary need is to get Boeing (through Tymshare/McDonnell-Douglas) to 
> >agree to release NLS/Augment to public domain. I think the best way to do 
> >that is for the CHM to appeal to Boeing that it should be released for 
> >historical, educational, and inspirational purposes. Because I expect it     (06)

I don't know about the "public domain" part.  Why not use something
like the GPL to guarantee that any derivatives remain free?  This
could also reassure the license owners that you aren't secretly
planning to make a bundle of money off something they overlooked.    (07)

> >In terms of hardware, I was able to get TOPS20 running (thanks to Ken's 
> >emulator) on a cheap $700 Linux box without any problems. Any standard 
> >desktop machine will be more than sufficient for our working team to get 
> >NLS running. We can discuss later in the project as to what kind of 
> >hardware would be best suited for the NLS that would be running at the CHM.    (08)

FWIW Doug's BI5 has been running on a relatively ancient Ultrasparc
which is outclassed by such $700 boxes.  However we are about to
transition it over to a dual processor 3.06GHz Xeon with 4GB RAM; they
haven't really played much with it yet, but speaking for myself I can
say the increased speed and memory makes it MUCH more pleasant to deal
with the otherwise lengthy and tedious steps needed for manipulating
lots of images.  They will also have much larger drives (IDE has
trumped SCSI).    (09)

My suggestion for a on-site museum exhibit would be to use something
fairly good, so people get fast and excellent response and aren't kept
waiting for their turn.    (010)

Oh, that reminds me!  If you want some historical veracity I will
happily donate the original Alpha 3000/400 used to develop the KL
emulator -- have kept every last bit of hardware, software, doc, and
boxes/packaging that DEC sent to me for this purpose.  Hint, hint,
hint.  It's a little slow by comparison with the Xeon, though :-)    (011)

> I'm particularly interested in what it would take way down stream to enable 
> multiple
> people to access the NLS/Augment system over the web as a "cyber exhibit". 
> That's
> a ways off, but it's interesting to start thinking it through.    (012)

For web access on the other hand it would probably be better to run
multiples of VERY cheap servers that aren't the latest/greatest but
allow one to easily increase or decrease their numbers as needed.  You
would also need to put all of them behind a firewall so it's not
necessary to worry about managing network security via TOPS-20 itself.
I've thought about this quite a bit in the past since we wanted to
bring up a "Public ITS" system hosted by Digex.  The plan was to have
multiple ITS systems (which have minimal security) behind a firewall
that only allowed trusted/registered users to come in, and permitted
no outbound connections.    (013)

We got the systems ready (actually had 4 of them on a single box) but
it never actually went public because I suddenly had kids and Digex
more or less went under when the bubble burst.    (014)

All these could be part of a much larger "Networking in the 70s"
protected ecosystem, but I'm digressing wildly.    (015)

> >>I worked on a TOPS-10 system running on a PDP-10 for five years
> >>back in the mid-1970s, but expect that I'm pretty rusty. Our favorite
> >>"game" was TICO roulette: you type in a random word and see what    (016)

I think you mean TECO :-)    (017)

--Ken    (018)
<Prev in Thread] Current Thread [Next in Thread>