Talk:UNIX System V

Pronounciation and System V history tree

Is it pronounced "System Vee" or "System Five"? Dismas 03:01, 2 August 2005 (UTC)

It is pronounced System 5. System 4, 3, 2, and 1 were earlier versions. Also, I'd like to point out that IRIX (A SVR4 with BSD extensions) is not mentioned in the article and is not present in the graphics of the UNIX tree.--RageX 04:01, 20 December 2005 (UTC)

Actually, there were no Systems 2 and 1, AFAIK. This is interesting though: does anyone know where the name System III came from? I have System III source code (available here), and it's documentation refers to it as UNIX Edition 3.0.

IRIX isn't in the tree because it isn't important enough. But maybe it should be mentioned in the article, in a list of System V derivatives? Qwertyus 14:16, 20 December 2005 (UTC)

IRIX is not important enough yet MINIX some how makes it in? How about a more accurate and large tree like so http://www.levenez.com/unix/history.html --RageX 08:11, 25 December 2005 (UTC)

IRIX, based on AT&T code, isn't important enough, but MINIX, which has no AT&T code, is? Methinks the original creator of the graph (it ain't a tree - it's not even connected, as there's a completely unconnected subgraph for, wait for it, MINIX) needs a visit from Mr. Cluebat. Guy Harris 08:33, 25 December 2005 (UTC)

HP-UX is more important than both and is not mentioned at all. to tell the truth, the article mentions SCO's products as being among the most used SYSV Unixes when they're actually irrelevant today with all the SCO's worries. is my opinion that the article should be edited to remove SCO's UnixWare and OpenServer from that paragraph and include HP-UX instead.CovardeAnonimo (talk) —Preceding comment was added at 00:18, 30 January 2008 (UTC)

I see your point. The problem, I think, is that this is actually a Unix history graph, rather than a System V tree; it isn't about "The many divergents of System V". If someone were to make a new, SysV-specific tree, then that should include IRIX. (Volunteers? :) Qwertyus 23:04, 25 December 2005 (UTC)

I think "System I" would have been "UNIX/TS 1.0", which, I think, was an AT&T-internal mixture of a V7 predecessor and some parts of PWB/UNIX. ("TS" was for "Time-Sharing"; there was also a "UNIX/RT" which was built atop a MERT base.). "System II" was probably PWB/UNIX 2.0, which added the rest of PWB/UNIX to UNIX/TS. "UNIX 3.0" was the result of merging that with "Columbus UNIX" (extensions developed by Bell Labs in Columbus, Ohio) a/k/a "CB UNIX" to make a somewhat unified UNIX. 3.0.1 added support for the PDP-11/44, and that was released outside of AT&T as "System III", rather than "UNIX 3.0.1", for some reason. Later 4.x releases weren't generally released outside of AT&T, but UNIX 5.0 was released as "System V". Guy Harris 08:07, 24 December 2005 (UTC)

The 1984 Bell Labs Technical Journal on UNIX, on pg. 1794, has a table of UNIX System versions with external equivalents which were used in the article for performance testing. Listed are: PG 1C-300 (1977 on PDP11) (no external release); UNIX 3.0 (1980 on PDP-11, VAX) (System III); UNIX 4.0 (1981 on PDP-11, VAX) (no external release); UNIX 4.1.1 (1981 on AT&T 3B20S) (no external release); UNIX 4.2 (1981 on PDP-11, VAX, 3B20S) (no external release); UNIX 5.0 (1982 on PDP-11, VAX, 3B20S) (System V). PG1C-300 refers to the internal release from the USG within Bell Labs (a footnote says 'The UNIX Support Group Generic 3 system is a derivative of AT&T Bell Labs Research Version 6'). That article is the only reference I've found with any slightly substantive information on UNIX 4. --Agarvin 22:48, 3 October 2006 (UTC)

Based on some discussion over time on The UNIX Heritage Society mailing list, a growing consensus has been that the numerical scheme employed for the commercial releases most closely relates to the Programmer's Workbench UNIX line. Most references can be found throughout https://www.tuhs.org and its associated mailing list, but if anyone picks up an interest in documenting this and wants further info, an email to that list will probably get to me:

UNIX initially starts in Murray Hill on the PDP-7 with the work of Ken Thompson, Dennis Ritchie, and a growing cast of other researchers, developers, etc. as time goes on. Well, one of the principle motivating factors for the Labs was the prospect of using UNIX as a typesetting system for patent applications. Promoting this project, the Murray Hill team was able to secure a then-new PDP-11 and quickly ported the fledgling system to it along with the typesetting system ROFF.

This secures UNIX its place as a formal tool used in the Bell System. Another early adopter was the Columbus, OH Bell Laboratories branch, which takes particular interest in UNIX as a basis for their Switching Control Center System. Their particular adoption was so thorough and specialized that this branch, CB-UNIX as it began to be called, shines out as a distinct internal branch of UNIX in the Bell System.

Well, as more Bell System units are looking to use UNIX, Bell Labs realizes they need a tier between the researchers in Murray Hill mostly working on exploratory and R&D work and the end users who are just engaged in day to day operations. Out of this need the UNIX Support Group (USG) is established to provide this buffer as well as engage in other QoL projects for UNIX users that can't be met by Murray Hill. Born out of this work is the "Program Generic" line of UNIX. My understanding is that Program Generic served two purposes:

- Program Generic UNIX was the "standard" UNIX for user applications in the Bell System

- Program Generic UNIX was the kicking off point for other branched UNIX projects such as MERT

The latter I can't completely verify, but the MERT 0 manual on TUHS indicates that the project was based on Program Generic Issue 3 (1976).

In any case, yet another group in Bell Laboratories began development of the Programmer's Workbench or PWB distribution of UNIX. As opposed to the USG Program Generic line, PWB was meant specifically for programming projects. One of the fuzzy bits of the history I don't know is whether PWB started with the research or USG codebases. Additionally, there may be some interplay between these and CB-UNIX, there are several examples of flow of ideas between the different teams.

Well, as of 1977, there are at least 5 major lineages of UNIX which have cropped up in the Bell System: Research, CB-UNIX, USG Program Generic, Programmer's Workbench, and MERT. The latter is a bit specialized of a case for virtualization work that was being done at Bell, MERT becomes DMERT which is the system 5ESS (on the 3B20D) is ultimately stood up on. As an aside, this system lineage may exist to this day in the emulated components of modern Alcatel-Lucent and Nokia-Bell Labs 5ESS stuff, but my understanding of that timeline is fuzzy.

So at this time, the Bell System is a very tightly regulated telecommunications monopoly that can't just willy-nilly start selling computers and software. Instead, the consent decree allowing their regulated monopoly stipulates that inventions must be provided to other industries at a reasonable fee, with the flip side being other industries having such liberal access to Bell System developments cannot use said developments to compete with them in the telecom industry. Given all of this, UNIX became very popular in the university scene since it was very accessible to educational institutions. One of the most important of these institutions is the University of California at Berkeley. Ken Thompson, visiting on sabbatical, works with folks at Berkeley to setup a V6 installation on a PDP-11 they have going there. Little did he know he was sewing the seeds for a project that would have lasting implications for UNIX. The extensions Berkeley developed on their system were eventually released as 1BSD and 2BSD. More on BSD in a moment.

So by the late 70s, Bell Labs is engaged in portability work, trying to move UNIX to the Interdata 8/32 and subsequently the VAX-11/780. Various Bell System Technical Journal articles and other historical research have indicated there was also work at this time on 8086 support as well as some fledgling work with the Bell System's own CPU architecture designs (BellMAC, 3B stuff, etc.) although the most important of these would be the VAX work. However, this is happening in research. Meanwhile, the other groups supporting UNIX in the Bell System are instead focused more on user-visible features, especially in the PWB lineage. As time goes on, the auxiliary branches of UNIX pull over bits and pieces of what research is working on, and occasionally a project from one of those branches will work its way into research, but the growing distance between PWB and research is becoming more and more significant. In 1979, new issues of CB-UNIX and PWB/UNIX are released in the Bell System, versions 2.1 and 2.0 respectively. As another aside, numbering between different UNIX lineages is largely meaningless, a CB-UNIX at 2.x and PWB/UNIX at 2.x reflect nothing of the specifics between them, those just happen to be the second issues from each project. In any case, around this time, yet another UNIX version pops onto the scene: 32V. 32V is the result of the VAX portability efforts and is essentially research version 7 with the portability additions. Berkeley takes a particular interest in 32V, especially in that it did not make full use of the VAX's virtual addressing capabilities. Given that, Berkeley spins up their first complete release, 3BSD, a modified 32V with virtual memory capabilities.

This is a bit of a shock to Bell. Before this, there hadn't been so major of a UNIX development outside of their shop that people went with that, rather than their own offerings, when procuring a UNIX system. The way the licensing worked, the Bell System was still getting their royalties, but often how it worked out is folks would negotiate and purchase a 32V-inclusive license from Western Electric and then go order BSD tapes from Berkeley. AT&T starts cooking up plans to market UNIX themselves and to get more involved in the market performance of their baby. Well at the time they start working on this, the two versions of UNIX most suited to selling to end users are the research line and Programmer's Workbench. Indeed, V7 included a host of changes from PWB as they were deemed useful to the userbase of the system as a whole. Bell folks settled on Programmer's Workbench as the basis for their commercial efforts, while pulling in improvements from other branches throughout the company to ensure it is solid. What this results in is PWB/UNIX 3.0(.1) serving as the basis for the first true commercial offering as an AT&T product (as opposed to Bell Labs just shipping tapes and some papers to parties with the licensing fee.) Unfortunately, AT&T up until this point had been a very tightly controlled telecommunications monopoly. Every aspect of their operation is concerned with telephones and the services that support them, they don't have the corporate machinery for marketing software, supporting a user-base, etc. so a lot of these matters weren't as tightly coordinated as from other shops like IBM and DEC. As an example, the manuals shipped for users of UNIX System III were really just back-stock of PWB/UNIX 3.0 manuals, as evidenced by their front page saying "UNIX Release 3.0" rather than "UNIX System III".

So System III drops and AT&T gets a bit more formal in the UNIX marketplace. Berkeley, not satisfied with 3BSD, continues tinkering on their own branch and begets 4BSD, which ticks up in popularity due to performance and usability improvements. Bell Labs, meanwhile, is internally starting to use PWB/UNIX 4.0 (which by 3.0 we know had dropped the PWB label, the story there allegedly is that management feared "Programmer's Workbench" would give users the impression the system was literally only for programmers.) This lineage at this time is also variably called USG UNIX or UNIX/TS, although the latter is actually the name of a very obscure branch of UNIX that I wouldn't do service by trying to describe in totality here. Anecdotally it was akin to the USG Program Generic project as an attempt to produce a "standard" UNIX base that would underlie future projects, but by that point it just got lost in the crowd, although it does have some influence on what comes after. In any case, I have nothing authoritative on UNIX/TS save that the lineage that begets System III and System V is *not* directly UNIX/TS, but incorporates components of it.

So 1981's UNIX 4.0, the main mystery at hand, is pretty much what one would expect, it represents a transitional period between 3.0 and 5.0. The familiar IPC interface of System V is largely there, the LP print system has started to replace the older lpr system, and platform support starts to grow with Bell porting UNIX to work on their then-new 3B20 machine architecture. One reference I can easily give is this: https://gitlab.com/segaloco/pwb4u_man This is a restoration of the 3B20S UNIX User's Manual Release 4.1, restored based on manpage sources from System III and System V in reference to a physical copy of said manual I have in my possession. Additionally here is the cover to said manual with a photo of the 3B20S: https://commons.m.wikimedia.org/wiki/File:UNIX4.1UsersManualCover.png

So all this while, the CB-UNIX branch has also been experiencing healthy maintenance and growth. In 1981, CB-UNIX Edition 2.3 is issued in Columbus, and at the same time, USG is considering their next commercial version of UNIX as well as working on their own next internal release. This would become UNIX Release 5.0, the last version of UNIX to use this numbering scheme. UNIX 5.0 incorporates the init system from CB-UNIX as well as several other improvements. Keeping with a Bell System policy at the time (which I can't verify, anecdotal evidence at best), only the odd numbered versions of USG's commercial product were to be released. This is given as one of the main reasons for 4.0 (or as it would've been called, System IV) never existing. UNIX 4.x certainly did exist and was in wide use throughout the Bell System, especially on 3B20 computers, but its maturation into UNIX 5.0 resulted in a system Bell decided to actually market. It's unknown what remnants, if any, of Program Generic-proper might have still existed. MERT and subsequent iterations were well entrenched in 5ESS by this point. PWB had amalgamated with other branch developments over time to become the commercial line. Research had successfully managed to keep its own branch of development going as well. Finally, there is BSD. Research eventually incorporates large portions of 4.1BSD before issuing a new version, V8, in the mid 80s. By this time, research and USG lineages have grown quite distant, with research much more closely resembling BSD (which in turn much more closely resembles V7 than the USG line does at this time.) The 80's march on, the workstation revolution, lawsuits happen, etc. and the story of little old UNIX 4.0 disappears into the sands of time.

Take this all with a grain of salt, I wasn't there, but I took the challenge of "What happened to UNIX System IV" years ago and this information is the result of a lot of that research.

Hopefully between the few links I've tossed up here someone will be able to get in touch or will reply to this talk article if there are any questions on this content. I'm hesitant to make the edits to the page myself largely because, again, this is the results of much of my own research. I'd rather this information filter through someone else vetting it before it winds up in the article. And also for anyone doing serious UNIX research, consider joining The UNIX Heritage Society mailing list, or at the very least be aware it exists, because it is an endless font of useful and novel information about the system as well as many aspects surrounding it. 2601:602:680:AE80:C8EF:84E5:F7A4:B371 (talk) 02:27, 20 September 2023 (UTC)

Good research. Some comments:

• I seem to remember something that gave me the impression that AT&T used the term "Generic" as a noun to refer to particular pieces of software; part of that may have come from interactions with the AT&T people when I was at Sun in the early days of the SVR4 project. Perhaps a "Generic" was an internal product that would be customized by individual users or for individual machines - it may have dated back to the software used on ESS systems.
• As for UNIX/TS, "The UNIX Operating System as a Base for Applications", from the July-Auggust 1978 issue of the Bell System Technical Journal, says, on page 2203 of the journal that

Improvements were made that allow UNIX processes to communicate and synchronize in real time more easily. However, additional real-time features were needed to control the scheduling and execution of processes. In another research organization, H. Lycklama and D. L. Bayer developed the MERT (Multi Environment Real-Time) opeating system. MERT supports the UNIX program-development and text-processing tools, while providing many real-time features. Some projects found MERT to be an attractive alternative to the UNIX system for their products. This led, in 1978, to the support of two versions of UNIX: the time-sharing version (UNIX/TS) and the real-time version (UNIX/RT). UNIX/TS is based on the research version of UNIX with additional features found to be useful in a time-sharing environment. UNIX/RT is based on MERT and is tailored to the needs of projects with real-time requirements. The centrally supported UNIX/TS will become the basis for PWB/UNIX as described by Dolotta et al., and will provide computation-center UNIX service. ...

The paper by Dolotta et al indicates that PWB/UNIX was originally based on Research UNIX; presumably the plan at that time was to switch to UNIX/TS as the basis for future releases. If the UNIX/TS described there isn't what ended up being UNIX/TS, that might be interesting to note; in any case, the flow of features from various UNIXes (Research, CB, PWB, etc.) into 3.0/S3 and successors would be interesting.

• UNIX 3.0/S3 had an /etc/inittab-and-run-levels-based init, as per UNIX User's Manual Release 3.0; does UNIX 5.0 incorporates the init system from CB-UNIX refer to additional init features and possibly the structure of what particular run levels mean? Guy Harris (talk) 07:46, 20 September 2023 (UTC)

system V

the new (last few days) release of Project Indiana is based on OpenSolaris, which is System V based (sun licenced it from Novell to replace sunos) —Preceding unsigned comment added by 86.16.160.17 (talk) 22:06, 6 May 2008 (UTC)

More accurately, Sun and AT&T jointly developed SVR4, before Novell's involvement with UNIX. For example, SVR4's virtual memory system and shared library mechanism, for example, were based on the SunOS 4.x VM system and shared library mechanism. Guy Harris (talk) 08:18, 20 September 2023 (UTC)

How did SV differ from UNIX 5.0?

What was in System V "Release 1" that wasn't in UNIX 5.0? (For that matter, what was in System III that wasn't in UNIX 3.0.1?) Guy Harris 23:33, 1 January 2006 (UTC)

Chart is not NPOV

Calling the operating system GNU/Linux is taking a POV in the GNU/Linux naming controversy.

Giving it any name is taking a POV, then!

Calling it anything other than an OSes based on the Linux kernel is very much taking a position on the matter as not all systems using Linux have any GNU code being used at all. Use of Linux does not require use of GNU. —Preceding unsigned comment added by 74.73.31.156 (talk) 06:20, 27 January 2008 (UTC)

I think you have it the wrong way around. I can think of systems using GNU that don't use Linux. GNU/hurd and GNU/Solaris come to mind. I cannot however think of any system that uses Linux that does not use GNU. —Preceding unsigned comment added by 202.1.119.216 (talk) 12:34, 27 January 2008 (UTC)

Change title to Unix System V?

I propose to change the title of this article to "Unix System V", as that is the capitalization used in this article, most of the Wikipedia, and the main articles Unix and Unix-like. OTOH, one could argue that SysV vendors have always (to my knowledge) called it UNIX (with the exception of Novell/SCO in the case of UnixWare), so I've decided to discuss the matter here first. Opinions, anyone? Qwertyus 16:00, 10 April 2006 (UTC)

I am in favor of renaming it to "Unix System V" because the lowercase version is what is used in the main Unix article. It just seems more natural and accepted to use the lowercase, though that is just mho. --Douglas Whitaker 17:21, 19 July 2006 (UTC)

According to X/Open's Trademark Usage Guide the proper usage is all-caps. When System V was published, all the brand names and literature used 'UNIX'. In Peter Salus's A Quarter Century of UNIX he discusses the capitalization controversy a bit and states his reasons for prefering Unix over UNIX; however, when discussing actual brands like System V, he still uses the all-caps version.

Images

I could provide some images of SCO System V/386 install media (3.5" floppies) and various manuals if it would be thought to contribute to the article. The diagram in the article is good, but I think that an actual image could contribute more to it (I just don't know of what though.) Any ideas?--Douglas Whitaker 17:21, 19 July 2006 (UTC)

SVR5

Is there any news to SVR5, or was it just something released for marketing purposes? —Preceding unsigned comment added by Frap (talk • contribs) 21:39, 20 October 2007 (UTC)

Yes, it was released as SCO UnixWare 7, but AFAIK never licensed to other distributors in the same way previous versions. I seem to remember a quote from someone at Sun saying they wouldn't licence it since the version number bump was just marketing, but haven't looked around to try finding it - we're talking ten years ago now. CrispMuncher (talk) 19:57, 13 July 2008 (UTC)

Linux is not Unix

GNU and/or Linux is a Unix-like system. It could be described as *nix, if desired. It has never undergone certification by the Open Group. Calling it a Unix implementation is factually incorrect. --ROM SPACEKNIGHT 02:14 2008-01-27 —Preceding comment was added at 02:15, 27 January 2008 (UTC)

Court filings

This article was submitted as an exhibit in the SCO v. Novell court case. [1], exhibit 2. --Reuben (talk) 16:50, 7 February 2008 (UTC)

The great SVR4.1 mystery

Many places on the web (including WP before my recent edit) claim that SVR4.1 introduced "asynchronous I/O" but none of them give sources. In fact no sources admit of the existence of SVR4.1 other than SVR4.1ES (Enhanced Security). I think I've solved the mystery - In "UNIX Filesystems: Evolution, Design, and Implementation" by Steve D. Pate, ISBN: 978-0-471-16483-8 we find various erroneous statements "UnixWare was based on SVR4.0", and also "SVR4.1 included asynchronous I/O. SVR4.2 included access control lists" (page 10). HughesJohn (talk) 22:04, 31 October 2008 (UTC)

Discontinued?

I just noticed a recent category category change from discontinued software to discontinued operating systems. I don't dispute the merits of that particular change but should it be categorized as discontinued at all. System V is still shipping. Sure, AT&T don't ship it anymore and it has changed ownership a few times but it is barely ten years since SVR5 was released, and that product is still being sold. Even if we don't acknowledge that as a reference version, it still seems inappropriate to tag this software as discontinued when it is the foundation of most commercial Unixes. CrispMuncher (talk) 14:48, 6 January 2009 (UTC)

I agree. Removing it from category discontinued since it isn't (yet). citation: http://www.sco.com/products/unixware714/ and http://www.sco.com/products/openserver6/ HughesJohn (talk) 14:05, 7 January 2009 (UTC)

the file system

Uh, i can't seem to find a suitable spot to mention the System V filesystem ; perhaps someone else could ? A few sources :

• file:///usr/src/linux/Documentation/filesystems/sysv-fs.txt (if you're on a Linux system)
• http://tldp.org/HOWTO/Filesystems-HOWTO-9.html#ss9.26

--Jerome Potts (talk) 11:26, 24 March 2010 (UTC)

SVR3 vs. SVR4 features

I distinctly remember that some of the features listed as introduced in SVR4 ware available in SVR3.2 based INTERACTIVE UNIX: TCP/IP stack, NFS, Sun RPC with XDR... (most were optional, though). Also, it would be nice if someone with handy reference material would mention the switch from simple getty to much more complex serial line terminal handling (details of which escape me at the moment :)). --bonzi (talk) 18:47, 2 May 2010 (UTC)

Unix history tree svg Image

The image in my computer doesn't see very well. The letters overlapped. --Solde9 (talk) 02:25, 30 July 2011 (UTC)

Market share compared to BSD systems (and other OSs)

This article seems to be missing some info on the economic side, and I've been staring at the following claim with growing surprise:

"System V was known for being the primary choice of manufacturers of large multiuser systems, in opposition to BSD's dominance of desktop workstations."

I'm sure SysV was popular for big systems (I remember UnixWare SVR4.2 running on ISP hosts, and AIX hosts serving networks of PCs), but wasn't IRIX a SysV variant for the desktop? Also, do PCs count as workstations? In that case, it's important to note that Xenix was the most popular Unix flavor of the 1980s, and that was a SysIII/SysV. QVVERTYVS (hm?) 21:40, 21 November 2013 (UTC)

Componentization

I removed the ESR quote

The SVr4.2 versions, due to malignant idiocy by USG, typically do not include on-line man pages.

from the article, because it doesn't really add anything but abuse. However, what is missing from the article at present, and what underlies the quote, is the gradual componentization of Unix by AT&T and others, where instead of getting the full system with sources like you got from BSD, you bought a "base system", then a "software construction system" (SCS, i.e. compiler + libraries + headers), a document processing system (troff + utilities), networking subsystem, etc., with prices obviously going up each time. QVVERTYVS (hm?) 10:22, 6 May 2014 (UTC)