Hacker News new | past | comments | ask | show | jobs | submit login
A bit of XENIX history (seefigure1.com)
75 points by luu on April 17, 2014 | hide | past | favorite | 35 comments

Hmm. I joined SCO in early 1991, and one of my first jobs in techpubs was working on documenting compatability between SCO-branded Xenix and SCO's release of SVR3.2 UNIX -- which was able to run binaries compiled for SCO Xenix (unsurprisingly) but offered a bunch of extras. AIUI SCO had been doing a lot of development of Xenix from 1986/87 onwards, when Microsoft made the strategic decision to focus on OS/2 and the successor to DOS. Taking on Xenix was what enabled SCO to grow to a $200M/year turnover multinational in about 5 years; and failing to understand the implications of Linux was probably what killed SCO (or rather, when they finally got it, they split the company and sold the UNIX IP to Caldera, who renamed themselves SCO and attempted to sue the universe) -- the rest is history.

I also worked for SCO (part of IXI in Cambridge) and think I even met you once :-)

The name SCO was very deliberate - the folks in Santa Cruz were a different company but when dealing with Microsoft people in Redmond wanted to seem like part of Microsoft. Consequently "Santa Cruz Organization" made it sound like they were a branch office rather than outsiders.

Many don't know what made SCO successful. Imagine you were a dentist. A local VAR would package up a system for you, such as a Compaq server, tape drive (for backup), terminals, SCO Unix, appointment and finance software and install it. SCO Unix tried really hard to be easy to use, ranging from the documentation to the software. (Remember it is a dental employee running backups not an IT specialist). The VAR made their money via a ~15% markup. There wasn't really much competition for small business systems at the time. Trivia: it used to be that the only software for managing a pet cemetery ran on SCO.

Caldera wanted SCO for access to that network of 15,000 VARs. The VARs were used to their markup on a $1k or more copy of SCO Unix, but it isn't as attractive as 15% of a $29.95 copy of Caldera Linux. Caldera would have been able to sell for more, but then the VARs would be questioning why they needed Caldera.

In the end it all failed, largely because Caldera wasn't needed. The irony is Clayton Christensen (of The Innovator's Dilemma) even spoke at the SCO annual event (SCO Forum) in 1999. SCO at the time was focussing more and more on "the enterprise" where Linux couldn't possibly be relevant :-) https://www.computerworld.com.au/article/51414/sco_forum_tec...

> Hmm. I joined SCO in early 1991

Meta question, and I'll try to ask it delicately:)

I know that you worked for SCO along time ago and its a different company than the one who started the lawsuit, but have you every had anyone refuse to hire you, or even joke about it, because you worked at SCO/had the name SCO on your resume?

With the SCO lawsuit a few years back there was an awful lot of vitriol from some people about how they'd never hire someone with SCO on their resume. I'd be curious to know if anyone ever did get black listed due to working at SCO/Caldera.

Ha! Charlie left the software industry and moved to bigger and better things (http://www.antipope.org/charlie/blog-static/fiction/faq.html). He does have some good stories about his days at SCO on his website (e.g. http://www.antipope.org/charlie/blog-static/2009/06/how_i_go...).

Interesting question, that ...

I was out of the industry by 2002 (do not ask me about what it was like to be part of a team making our first pitch for VC seed investments on September 12th, 2001 ...). But I am informed by a barrister of my acquaintance (over a pint) that because of a very interesting precedent set in English employment law following the collapse of BCCI in 1992, if SCO were still in business in the UK, and I was still in the field and had difficulty getting work because of the aforementioned vitriol, I could sue them for damages to my reputation and have a good chance of collecting.

(The Bank of Credit and Commerce International, BCCI, collapsed in an epic scandal of corruption, fraud, and money laundering: innocent former employees of BCCI subsequently sued the liquidators and received damages in compensation for the trauma and injury to their reputation, which in many cases rendered them unemployable in their industry thereafter. More on BCCI here: https://en.wikipedia.org/wiki/Bank_of_Credit_and_Commerce_In... )

cstross has commented that as a result of The SCO Group's actions, he considers the association in his resume to be a liability. Lessee ... here we are:


(Which reminds me: according to a legal friend of mine, here in the UK it's actually a civil offense for a corporation to bring its name into such disrepute that it damages former employees' prospects. This harks back to the BCCI collapse in the early 90s, but suggests that if I was still in the software biz I might actually have a case in law against post-takeover SCO for shitting all over my resumé. But I digress ...)

(Yes, the whole bit is a parenthetical.)

I've only somewhat recently discovered Stross's fiction and blog, I recommend both strongly.

The cut off seemed to be more about people who stayed after the lawsuits started, rather than anyone who had ever worked there. That is a significantly smaller number of people. I haven't heard of any people having problems. These days the reaction I get on mentioning SCO is "who?".

The company I started out at wrote the (infamous) C2 subsystem for SCO Unix. Every once in a great while, some elderly neckbeard will give me shit about that. Memories...

XENIX was my first nix. Back around 1985 my high school had a Tandy TRS80 with the 68000 processor and 6-12 terminals. They used it to replace a Burroughs mini for the COBOL class. I just barely missed punching cards for the Burroughs and instead learned vi and stupid tricks like writing to other users' ttys. I still torment my Microsoftie friends when I remind them that they're the ones that got me started on nix, long before Linux or OSX came around.

My next phase of using XENIX unfortunately never happened. I ran some version of MINIX on 8088 hardware and would gladly have picked up XENIX if it were available sub-$100 back then. Instead I migrated to DESQview, which amazed me with what it was able to accomplish compared to early Microsoft Windows versions. It wasn't until 386BSD appeared that I ran a unix on home PC hardware again before moving on to the early Linuxes.

The TRS-80 Model II/16 or just the Model 16?

It shouldn't surprise you that there was also a version of 68000 Xenix for the Lisa.

Not sure. I'm also one of the few that used A/UX on the 68k Macs.

I was Intel at the time and when the first pieces of the 80386 came back from the fab functional Intel was scrambling for ways to test it (I was in Systems Validation). We tried running Xenix on it and discovered various places where Microsoft's kernel team had 'cheated' by using reserved bits in various places which just happened to work on the 286. Needless to say it caused quite a commotion :-)

Remember the 32-bit multiply bug?

Wow, the infamous "merged product"... In '89 I joined MS to run the "LM/X" product, where we extended the Unix networking model to support the SMB protocols. A very interesting time where MS worked with HP to get the kernel and networking changes into the mainline code base while at the same time AT&T was doing their own version. Sun was pushing their implementation of PC NFS, and many interesting "standards" meetings were held. If I remember correctly, the PC networking protocols showed in in XPG4 (X/Open Portability Guide) in 1992.

I wonder what would have happened if Xenix was the successor to DOS. We would probably be living in a completely different computing world today.

Hardware was too expensive. By the time RAM/HD/CPU prices became low enough to store and run a UNIX OS, something else would have came along and taken it's place.

You could run DOS on a $999 computer, and you could run Windows on a $1999 computer. It took a $4999 computer to run UNIX. That's basically what kept home users on such futile operating systems. That's also what kept Windows NT off peoples desktops until 2000 or so.

I disagree. I remember running Minix on a no-name 8088 machine with 640K of RAM and no hard drive at all. Unix was developed on machines with far less resources than that. It is not a particularly heavy-weight OS.

Sure, Minix would run on a low-end system, but to have taken the place of Microsoft Windows as the successor to DOS you would have needed a GUI like X which Minix did not have.

And Minix was essentially a clone of IBM PC-IX, which was essentially a UNIX Version 7 for the PC and PC-XT that could also run off floppies.

In fact, MS did make Xenix for 8086 too.

Unix, schmunix. With a 7 Mhz processor and 256k RAM, the Amiga delivered multitasking, a GUI, interprocess communication, hardware accelerated graphics and a whole heap of other features in 1985 that the MS/Intel world didn't deliver for another 10 years.

CPU, RAM etc were not the constraining factors here.

I am suspicious of this claim.

It's tempting to imagine the "what might have been" had the first Win1 APIs been delivered on Xenix rather than DOS.

At least there was nothing unethical in this one unlike the later OS/2 2.x fiasco.

I have my doubts M$ was really that keen on that path; after all, there's a lot of stuff in Xenix that M$ wouldn't own and that's always been a problem for Gates and the boys. So OS/2 for a while was the heir apparent, which would have also changed things quite a bit (for good, I think, but many disagree). What we got was stepchild of VMS, which also could have been great, but has wandered off in it's own direction.

The possibility of running Xenix-native binaries elsewhere would negate much of Microsoft's later strategy. Hardware could be a commodity, but software should always be their own.

> Xenix should be the 16-bit “successor” to DOS

Shouldn't this be "32-bit"? MS-DOS itself was always 16-bit.

At the time, DOS and the original (8088) IBM PC primarily competed with 8-bit micros: 8080 systems running CP/M and the 6502-based Apple II.

In that context, they were probably talking about the upcoming generation of microcomputers, based around bigger chips: 80286, 68k, etc

The 8080 had a 16bit address bus, though; which is what I always assumed was meant by referring to a machine by it's "bit count."

By that logic, the 6502 is a 16-bit CPU, because it (and just about every other CPU we call 8-bit) has a 16-bit address bus.

The 68000 has a 16-bit data bus and ALU, a 24-bit address bus, and 32-bit registers, but is usually considered a 16-bit CPU. I guess for older CPUs, the data bus and the ALU size are the determining factors.

Xenix ran on the 8086, FWIW. As I recall, the intent of the statement was the thought on M$ part that users would want a more capable system when the new processors were mainstream and DOS would be retired and replaced with Xenix. Obviously, this didn't pan out.

We had the the SCO Xenix 2.3.4 running the billing system for the water and sewerage company on my hometown (about 100k homes) till the 2000s. The software was written in SGBD (FoxPro-like DB).

It was up and running till 1997 on a i386 with 16MB RAM and 1gb SCSI serving dozens of users in various locations via VT100. No TCP/IP stack running, it was not supported by our hardware.

I bet lots of companies still has this kind of setup, at least I saw a retailer doing the same setup but running MF Cobol instead of SGBD.

I worked for a company, around 2002, that offered a managed billing platform for small mobile-phone operators.

That platform was largely built around microfocus COBOL.

That was a blast for from the past - I spent 1986-91 designing serial cards and writing drivers for Xenix systems at Specialix. We had 32 terminals running on 386 Xenix systems, tell that to the kids of today with their Ghz CPU's and fancy VM's and they won't believe you.

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact