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)

ESS Switching releases were called Generics. Also Unix "System" releases were generally commercial and "Version" releases were generally academic. This entire article has lots and lots of inaccuracies, chronological errors, and just outright misdirection. For example System III came before Version 7 and System V after.

Unix was never shared with anyone for commercial use outside of AT&T until they were effectively forced to do so to settle a lawsuit with OSF (Open Software Foundation) which represented IBM, DEC, and HP (and others), the suit was filed because someone within the US Air Force Standardized Systems Group read the SVID (System V Interface definition) that came packaged with the 3B2/600Gs that had been sent to them by AT&T Government Systems and made the stipulation that in order to bid on contracts with the USAF Unix based computers needed to comply with that specification. This meant that anyone whose distribution was based on the academic code could not bid on those contracts.

Xenix, Sun-OS, DEC Ultrix and other Unix variations that escaped out in the wild came from academic code, not official AT&T commercial code. Sun Solaris WAS AT&T System V commercial code (well actually Bell Labs R&D Unix code) and was given to Sun as AT&T had a substantial stake in the company at the time and had ported it for their own use in 5ESS switch development. Every other System V based port was done by the manufactures using the 3B2 DEMon porting environment (IBM AIX, HP-UX, DEC Tru64, Pyramid, Data General, etc..).

I worked for Bell Labs (Indian Hill), AT&T Network Systems and AT&T Information Systems (ATT-IS) and saw it from all 3 perspectives as well as the interactions with Teletype, which to fully understand how all that came about you had to. After divestiture all AT&T divisions had to interact with each other as if they were all separate independent companies, particularly for anything having to do with commercial computing. It is a stipulation of the breakup called "Computer Inquiry 3". So System V Unix came from Bell Labs, was distributed by Network Systems, to run on hardware manufactured by Teletype and marketed by ATT-IS and they all had to deal with each other as if they were separate companies with separate book keeping and were very restricted as to what information they were allowed to share with each other.

So in many ways this article is overly complicated, and in other ways nowhere near detailed enough to paint the true picture. There is also a lot of things that are included as if they were standard that were not. The base distribution came on (23) 720K (Dual Density) floppies but a full suite with all the options was hundreds. I don't even know where I would start editing, as there is inaccuracies or misinformation in almost every paragraph including a complete omission of the ATT-IS/NCR merger, which is what really dictated the ultimate fate of the commercial distribution of the what is really the AT&T System V code base.

The System V code went from being nowhere outside of AT&T, except as a reference source that you couldn't legally change (at a cost of $80,000yr), to being everywhere, in dozens of independent interpretations, in a couple of years. The 3B2 code base was very specific to that architecture, the Western Electric WE32X00 family, a chip designed very specifically to run that OS. Essentially all ports were heavily adapted to whatever the porting organization needed it to run on, and done with a methodology inconsistent with that of Bell Labs. Even Solaris which was ported by Bell Labs itself is fundamentally quite different from the 3B2 base (WE32X00 vs Motorola 68020 and eventually SPARC). Skorpioskorpio (talk) 08:50, 1 August 2024 (UTC)

Unix was never shared with anyone for commercial use outside of AT&T until they were effectively forced to do so to settle a lawsuit with OSF (Open Software Foundation) Given that the OSF was founded in 1988, according to the Open Software Foundation page, either 1) that page is incorrect, 2) the lawsuit wasn't filed by the OSF, or 3) Unix was never shared with anyone for commercial use outside of AT&T before 1988, at the earliest. If 3) is the case, then since, according to UNIX System V, SV was first released in 1983, either 1) that page is wrong about the date or 2) that 1983 release, and all subsequent pre-1988 releases, don't constitute "[sharing Unix] with anyone for commercial use outside of AT&T". Which of those are the case?

(Presumably, "[sharing] with anyone for commercial use outside of AT&T" doesn't include somebody taking the source to a Research UNIX release, or modification thereof such as a BSD release, and licensing it commercially.)

Sun Solaris WAS AT&T System V commercial code (well actually Bell Labs R&D Unix code) and was given to Sun System V Release 4 was a joint AT&T/Sun project. The virtual memory subsystem and VFS layer of SVR4 were based, respectively, on the SunOS 4.x VM system and VFS layer, for example.

Even Solaris which was ported by Bell Labs itself is fundamentally quite different from the 3B2 base I"m not sure what you mean "ported by Bell Labs itself". Most of the SVR4 code was portable, all the way down to the kernel code. I was in the OS group at Sun between 1985 and 1988, and was one of the people talking to the AT&T people for the Sun/AT&T deal. AT&T appeared to have completely separate code bases for each platform on which SV ran; Sun had a single code base, including the kernel, that ran on Sun-2 68010-based machines, Sun-3 68020-based machines, Sun-4 SPARC-based machines, and VAX (an 11/780 at Sun used to make sure SunOS worked on it), up through SunOS 3.x. SunOS 4.0 completely rewrote the VM system, mainly to support memory-mapped files, but also to push the platform dependency in the VM system into the HAT (Hardware Address Translation) layer, meaning the rest of the VM system, and the kernel code that used the VM system, needed few if any #ifdefs etc. to deal with different memory management units. (The internal VAX port was dropped for 4.0; I guess the effort of writing a HAT ayer for the VAX was not deemed worth it.) SunOS 4.x's code base added support for the 80386, using its on-chip MMU, to support the Sun386i, for the 68030, using its on-chip MMU rather than the very-different Sun MMU, to support the Sun-3x models, and for SPARC processors with SPARC Reference MMU implementations, to support the SuperSPARC and hyperSPARC-based machines.

Presumably somebody in AT&T wrote a HAT layer for the WE32K processors, although part of the AT&T/Sun joint venture was to make SPARC a standard for UNIX systems, with the WE32Ks being supplanted by SPARC. (The idea was that SPARC+SVR4 would replace DOSTel and Wintel.) ("NARRATOR: this did not happen.")

(WE32X00 vs Motorola 68020 and eventually SPARC). I don't remember whether Sun ported SunOS 5/Solaris 2 to any Sun-3 (68k) machines; I don't believe they did. If there was an SVR4 source version for 68k from AT&T, I don't know whether they picked up any of Sun's 68k support from SunOS 4.x, or wrote their own. Similarly, I don't think Sun did a SunOS 5/Solaris 2 port to the Sun386i; there was, as I remember, an SVR4 x86 port from AT&T, using IBM-compatible PCs as the platform (as I remember it, rather than the WE32k/3B2 port, may have become the "reference" platform version for SVR4 at some point); I don't know whether AT&T picked up the usable parts of the Sun386i port, such as the HAT layer, or wrote their own. At least some other parts of the Sun386i port wouldn't be useable for a PC port, as the 386i was somewhat more like an 80386-based Sun workstation than like an IBM-compatible PC.

In any case, the SunOS 5/Solaris 2 code base in Sun was common to both the SPARC-baed Sun machines and the x86/IBM-compatible PC port of SunOS 5/Solaris 2, even if AT&T didn't pick up the "one code base for all platforms, with CPU-specific and platform-specific subdirectories as necessary" idea from Sun. (That model is how most UN*Xes work these days; in addition to Solaris, various components of Linux including the kernel and the C library, the BSDs, and Darwin and the Darwin-based OSes such as macOS, iOS, iPadOS, etc. all work that way.)

hardware manufactured by Teletype I think you mean Western Electric (the "WE" in "WE32k"), not Teletype. Guy Harris (talk) 13:13, 25 August 2024 (UTC)

OK, no idea how to do inline commenting in this forum, so:

Unix was never shared with anyone for commercial use outside of AT&T...

For clarity here, System V was a commercial product from AT&T and only AT&T (before OSF, you reference 1988, sounds about right) .. well major competitors I should say. There was a System V on Intel (i386) done by Lachmann & Associates, but it's more of an interesting side note than anything else. Most of Lachmanns business was providing consultants to Bell Labs and even their Naperville, IL headquarters were on land owned by Bell Labs.

Point being AT&T did not allow anyone to modify and redistribute their System V code until they were forced to do so. A source code license got you the ability to view the code for reference. Technically you were in breach if even modified it for your internal use, and you were certainly not allowed to modify, rebrand and distribute it.

BSD is not System V, it's derived from Version 7, it's academic Unix, not commercial Unix.

Anyway, your reply seems, at casual glance to be correct , but contains assumptions that this somehow negates dates, but it doesn't. Maybe I wasn't clear, System V was distributed by AT&T until it was shared as a result of OSF, nobody else had the right to distribute the System V code before that unless they were specifically given that right by AT&T and in conjunction with AT&T, as is the case with Sun Microsystems, Lachmann, Convergent Technologies (3B1, though that technically wasn't actually System V), and there was a few others for specific niche markets, like Olivetti, Xerox, Amdahl, and IBM (which was confined to System V UTS on mainframes and has nothing to do with AIX), oh and Digital Equipment Corp (Sort of).

Sun Solaris WAS AT&T System V commercial code (well actually Bell Labs R&D Unix code) and was given to Sun...

Your reply here pertains to the development of a commercially viable product, System V (R&D Unix) existed in telephone switching development and test labs before that, while Sun did assist in that effort it initially existed in very much a non-Sunified variant. No NFS (used RFS instead), no domaining for email (was all ball bat UUCP addressing), no NIS, no SunView, used ineternally developed StarLAN networking SBUS cards. And it was all SUN3 stuff, SPARCs weren't a thing yet.

Even Solaris which was ported by Bell Labs itself is fundamentally quite different from the 3B2 base...

In this section you mention; "I was in the OS group at Sun between 1985 and 1988, and was one of the people talking to the AT&T people for the Sun/AT&T deal.". While I don't recall the name, it is entirely likely that we've probably spoken in the past then. I was in the group that did the initial deployment of Sun3s into the 5ESS mega lab. one of the primary objectives of SVR2 was to get it to a single code base that was based on something commercially viable, initially 3B5/3B15 and pretty quickly to the 3B2/DEMON platform (WE32100) which consists of a 3B2 running SVR2 and a second 3B2 running DEbug MONitor firmware to develop low level code.This was around SVR2.1ish. SVR3 was mostly focused on networking, RFS, port forwarding, StarLAN, Dot Mapped Displays (DMDs), as well as laying the groundwork for voice integration via ISDN/PSDN.

Also I wasn't aware there even was a native Unix of any kind that ran on the 386 SBUS co-processor card, and not sure why that would even be desirable. It wasn't a very impressive performer even by the PC standards of the day, which in and of themselves weren't great compared to Motorola, NAS, MIPS or the WE32100, let alone the soon to be flood of RISC chips.

If there was ever any talk about migrating the SVR porting base to SPARC, that was after my time at Bell Labs. And I can't really imagine how that conversation would have come about. AT&T dumped it's stake in Sun around the time the SPARC was being rolled out, assumedly to make the NCR merger cleaner. Once AT&T and Sun were no longer joined at the hip, and the OSF stuff dealt with, Sun became not much different than the likes of HP, DEC, IBM and the rest of the OSF beneficiaries in the preferences it received from AT&T. Although Sun did have the distinction of having leftover R&D Unix bits in there, like David Korn's experimental tools version of KSH, which doesn't exactly work like it's supposed to.

Anyway, I went from there to Motorola just before SPARC made it's splash... That Motorola was none to pleased with. Motorola was pissed off about the whole thing enough to essentially develop it's own Moto-style Sun4 clone, or what they thought the Sun4 should have been. A 68040/VME bus machine that frankly probably sold more units inside Motorola than outside. It was not a success.

hardware manufactured by Teletype I think you mean Western Electric...

Nope, I meant Teletype, Western Electric was forbidden from manufacturing computer components, as part of "Computer Inquiry 3". The WE in WE32000 predates divestiture, and yes it stands for "Western Electic", but no data sheets on the product line say that. Most of that stuff was made at Teletype Skokie Works. Most of Western Electric, as a manufacture, got absorbed into AT&T Network Systems (if it was telephone switch related), Teletype (Data terminal and computer related), Consumer Products or Business Systems (if endpoint related), Cellular Infrastucture, or simply spun off to be used by the Baby Bells via BellCore. Divestiture pretty much gutted what was Western Electric, as it was always viewed as the problem, and the target of essentially every chunk that had been broken off of AT&T in every prior anti-trust suit. Skorpioskorpio (talk) 16:15, 25 August 2024 (UTC)

There was a System V on Intel (i386) done by Lachmann & Associates Is that the System V/386 described by this AT&T product manual from 1988? Point being AT&T did not allow anyone to modify and redistribute their System V code until they were forced to do so. As I remember, at least for System III, AT&T didn't sell it as a binary package to run on a PDP-11, they sold it as source code, with no support, under terms along the lines of what they offered for the V7 UNIX from Research, wherein the purchaser of the source code license could resell binary products made from it, with per-copy royalties paid by the reseller to AT&T. HP speaks, in their documentation of HP-UX from 1983", as having components coming from System III; see, for example, the init man page, which describes an S3-style init, with multiple run levels and an /etc/inittab file. The "HP-UX Compatibility Model" section of the manual says that

HP-UX is Bell System III plus "HP value added". HP value added includes both Hewlett-Packard capabilities, such as graphics, and features from other UNIX systems, such as those from University of California at Berkeley.

As of 1986, the HP-UX documentation, such as Volume 1A, Section 1 of the reference documentation, speaks of HP-UX as being "HP-UX is AT&T System V plus "HP value added".", in a form that sounds like the result of 1,$s/System III/System V/g.

As I remember, the "you can't change anything" came later in the lifetime of System V, around the same time as, or a bit later than, the System V Interface Definition.

System V (R&D Unix) existed in telephone switching development and test labs before that, while Sun did assist in that effort it initially existed in very much a non-Sunified variant I remember hearing something about that while at Sun, but that was an internal-to-AT&T project unrelated either to SunOS or to the SVR4 project.

used ineternally developed StarLAN networking SBUS cards. And it was all SUN3 stuff, SPARCs weren't a thing yet. Presumably by "SBUS" you mean something other than Sun's SBus, as that was introduced in 1989, on the SPARCstation 1, and did not exist on any Sun-3 hardware - or even on the earlier Sun-4 SPARC hardware, which first came out in 1987.

Also I wasn't aware there even was a native Unix of any kind that ran on the 386 SBUS co-processor card I never spoke of any such native UNIX. There was a native UNIX - SunOS 4.0.x - that ran on the Sun386i, which was a Sun workstation using the 80386; it came out in 1988.

If there was ever any talk about migrating the SVR porting base to SPARC, that was after my time at Bell Labs. As I remember, it was talked about during my time at Sun. Bill Joy presented the plan within Sun as SVR4+SPARC becoming the next standard platform, replacing the DOS/Windows+x86-based PC platform. (As noted, he... didn't get that one right.)

AT&T dumped it's stake in Sun around the time the SPARC was being rolled out, assumedly to make the NCR merger cleaner. At least according to Sun Microsystems § Operating systems, "in 1988 [AT&T] announced they would purchase up to a 20% stake in Sun." That was after SPARC came out. The article goes on to say that "By the mid-1990s, the ensuing Unix wars had largely subsided, AT&T had sold off their Unix interests, and the relationship between the two companies was significantly reduced." According to AT&T Corporation § The Break-up, "AT&T acquired NCR Corporation in 1991."

Anyway, I went from there to Motorola just before SPARC made it's splash... So that would be in 1987 or earlier, if by "SPARC made it's splash" you mean "SPARC was introduced". That would either have been before, or in the very early days of, the AT&T/Sun deal.

Nope, I meant Teletype, Western Electric was forbidden from manufacturing computer components, as part of "Computer Inquiry 3". They don't appear to have been forbidden from manufacturing computer systems for internal use. According to 3B series computers § 3B high-availability processors, the 3B20D was "first produced in the late 1970s at the Western Electric factory in Lisle, Illinois", so they were making devices that both AT&T and the US government may have pretended weren't computers, but that really were; 3B series computers § 3B20S says that the 3B20S "was developed at Bell Labs and produced by Western Electric in 1982 for general purpose internal Bell System use.", so pre-breakup. The AT&T fabs that made the WE32000 - or Bellmac 32, to use its original name - might not have been part of Western Electric, but I'd need a citation to believe that they were part of Teletype rather than an internal part of AT&T.

AT&T started selling computers and microelectronics post-breakup, starting in 1984. The chips came from what was called "AT&T Technologies" at that point, as per the WE 32100 Microprocessor Information Manual.

References

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)