I've assembled some notes from old manuals and other sources
on the formats used for on-disk file systems through the
Additional notes, comments on style, and whatnot are welcome.
(It may be sensible to send anything in the last two categories
directly to me, rather than to the whole list.)
> post-V7 AT&T
> Unices have always been unavailable, at least at any price that I can
> afford on my college student budget. Solaris, however, at least
> started out as an implementation of SVR4 and is freely available. How
> much of System V still lurks inside Solaris 10 (the last version to
> include such traditional workstation elements as CDE and DPS in the X
> server) and how much has been removed in favor of a more GNU-ish
> userland experience? Is Solaris a good way to get a System V
> experience without breaking either the bank or copyright law, or is
> this a hopeless situation?
I purchased a Solaris 8 source media kit from Sun for around $100 back
when they offered it. A lot of files still have headers saying
Proprietary Unpublished AT&T source code or some such. I don't know
how different it is from System V, because I've never been able to
procure a copy of that source. It sure looks like the Solaris 8
userland, at least, is almost pure System V.
Tim Bradshaw <tfb(a)tfeb.org> wrote:
> On 28 May 2009, at 18:00, Michael Kerpan wrote:
> > Solaris, however, at least
> > started out as an implementation of SVR4 and is freely available.
> > How much of System V still lurks inside Solaris 10 (the last
> > version to include such traditional workstation elements as CDE and
> > DPS in the X server) and how much has been removed in favor of a
> > more GNU-ish userland experience? Is Solaris a good way to get a
> > System V experience without breaking either the bank or copyright
> > law, or is this a hopeless situation?
> I never (other than transiently, and even then in various heavily
> bastardised versions such as Masscomps' RTU) used a Sys V Unix other
> than Solaris. However I did live through the SunOS 4 -> Solaris
> transition. My memory of that is that the early Solaris versions
> (2.2?) seemed extremely austere and unpleasant compared with BSD-
> derived systems. Solaris doesn't seem like that now, and in fact when
> I play with BSD derivatives they seem quite austere.
> So I would suspect that, no, Solaris is not any kind of good
> representative of what System V was once like. It's not a GNUoid
> userland (who knows what the next release will be like, if there is
> one? OpenSolaris seems to have drifted rapidly off into optimize-the-
> desktop neverland and I hope will not be representative of what the
> next Solaris looks like), but it's no more representative of what
> things were once like than any system still under development is
> representative of what its distant ancestore were like.
> (CDE is not a traditional workstation element in any real sense - it's
> pretty recent. I don't think it even existed in the early Solaris 2
To my knowledge, System V is not quite a well defined concept: there
were several releases and each had different features and capabilities.
The closest you can get today to test the flavour of running an early
System V (not a System V Release 4) native on real and affordable/common
hardware, I think would be getting some old Xenix for the 386 (SVR3),
which is floating around some P2P networks (in eMule I know it is).
Legality of getting it this way for personal and non-commercial use?
Depends on your local law, so beware. This Xenix which is on eMule lacks
TCP/IP (it was an add-on package sold separately), lacks the Development
Kit (compiler, headers, etc.; also sold separately), and obviously lacks
the source code. So if you already know Unix, in two days of use maximum
you should have seen all there is to it and be really bored about it.
The Development and Streams/Tcpip kits for this Xenix are also
"available" in some Internet "places", but I have not ventured to try
them because Xenix is a real bitch to live with...
If you want to inspect the source of a somewhat early System V Release
4, in eMule you also have a tar.bz2 file with the sources for Solaris
2.6 (beware, it won't unzip properly on Windows because some files names
are incompatible with Windows filesystem, i.e. "con" filename, etc.).
You won't be able to "run it", but you can inspect the source of a SVR4
which is only 8 years separated from the original SVR4 AT&T release.
And don't forget OpenServer 5.x, which is a UNIX System V Release 3.2
with many custom add-ons. Version 5.0.7 is still supported by it's now
bankrupt mother company, runs on modern Intel PCs, and has the really
old and rusty taste of ancient UNIX.
After trying all that, you would be thankfull for having the GNU userland
and the Linux kernel. That was my bottom line, anyway.
As a long-time Linux user and a long time Unix history buff, I've been
wanting to "play" with classic Unix variants for quite some time.
Obviously, Research Unix up through V7 and the BSDs are readily
available and I've at least mucked around with them, but post-V7 AT&T
Unices have always been unavailable, at least at any price that I can
afford on my college student budget. Solaris, however, at least
started out as an implementation of SVR4 and is freely available. How
much of System V still lurks inside Solaris 10 (the last version to
include such traditional workstation elements as CDE and DPS in the X
server) and how much has been removed in favor of a more GNU-ish
userland experience? Is Solaris a good way to get a System V
experience without breaking either the bank or copyright law, or is
this a hopeless situation?
I seem to remember that for System V TCP/IP that you needed
STREAMS first, so that was SVR3. It may have been back potred
but I don't know. And after that, then you had a choice of
Wollongong or Lachman implementations. Bell Labs had their own as well,
but I believe that was only available internally. Amdahl UTS used
Lachman (which I kind of remember might have been Convergent's code),
but at Indian Hill it was removed and the home grown one put in.
I don't know who did the kernel code, but the user land utilities
were BSD ports done by Ralph Knag in Murray Hill. This was an
interesting setup as it was System/370 hardware so it had a
channel to ethernet device from Spartacus, probably a K200 since
there was a "k200" command to fiddle with it. I largely ignored
TCP/IP initially as on the first UTS release, just telneting out of it
used a ton of system CPU, something would loop in the kernel instead of
going to sleep. Besides we had Datakit for interactive connectivity,
and NSC HyperChannel for intra-datacenter file transfer
(which I remember being something like 50mbs in 1987)
For the original SVR4, the official porting base was the 3b2 and that
group from Summit (which was later spun off as Unix Systems Laboraties
(USL)) used Lachman as well for it's TCP/IP.
> I was wondering if anyone had access to any SYSV for the VAX and could
> say what levels support TCP/IP?
> I put in a request at http://www.novell.com/licensing/ntap/legal.html
> to see if they are even entertaining the sale of SYSV licenses... But
> I kind of figure they don't have the actual material itself....
> I know A/UX a SYSVr2.2 had TCP/IP but I don't know if it was in the
> AT&T base, or if it was something that UniSoft had added...
> Anyways thanks for any/all responses....
> Oh and FWIW I've gotten a super minimal SYSIII thing booting on SIMH!
> I've just have to work out some more disk formatting/restoring as the
> root partition sizes don't agree between 32v & SYSIII....
(Sorry this was supposed to go out on my fairhaven account not my work email.
I do apologies if it comes through twice, still learning kmail!!)
Just out of curiosity, why not host some of the stuff on the tuhs website ? (i
could be wrong and their might be some copy write stuff but if dmr wont
On Wednesday 27 May 2009 05:25, Seth Morabito wrote:
> On Mon, May 25, 2009 at 7:33 PM, Jason Stevens <neozeed(a)gmail.com> wrote:
> > I don't know if it's worth mentioning...
> > but it looks like Dennis Ritchie's page is down....
> > http://www.cs.bell-labs.com/who/dmr/
> > At least there is the wayback machine
> > http://web.archive.org/web/20070930200555/http://www.cs.bell-labs.com/who
> Sadly, the Wayback Machine is now not serving up the page either:
> "We're sorry, access to http://www.cs.bell-labs.com/who/dmr/ has been
> blocked by the site owner via robots.txt."
> Indeed, the host's robots.txt file has this entry in it toward the bottom:
> User-agent: *
> Disallow: /
> so I assume that the site has been re-scraped since coming back up,
> and is now no longer made available by the Internet Archive, according
> to their stated policy on robots.txt exclusions.
> Just an oversight, I'm sure, but it shows off the fragility of
> information on the web. You cannot trust the Internet Archive to make
> information publicly available forever.
> TUHS mailing list
I was wondering if anyone had access to any SYSV for the VAX and could
say what levels support TCP/IP?
I put in a request at http://www.novell.com/licensing/ntap/legal.html
to see if they are even entertaining the sale of SYSV licenses... But
I kind of figure they don't have the actual material itself....
I know A/UX a SYSVr2.2 had TCP/IP but I don't know if it was in the
AT&T base, or if it was something that UniSoft had added...
Anyways thanks for any/all responses....
Oh and FWIW I've gotten a super minimal SYSIII thing booting on SIMH!
I've just have to work out some more disk formatting/restoring as the
root partition sizes don't agree between 32v & SYSIII....