Enterprise Unix Roundup: Ghosts of Xenix Past

Microsoft's newfound focus on intellectual property enforcement and other recent news developments make a nice launch pad for conspiracy theories about whether ties that bind link SCO, Xenix and Linux?

Various theories are circulating as to why Microsoft has been so gung-ho about intellectual property enforcement lately. I have opined about it myself in other venues, basically putting forth the theory that the staff at Redmond is running a bit scared right now as they watch Linux slowly encompass the server market.

Other pundits have tossed out a very interesting theory this week, as the Commonwealth of Massachusetts called for public comment regarding the inclusion of Microsoft's Open XML (OOXML) document format as an "open" standard Massachusetts can use for government documents. In light of these events, there's a pretty good argument that all of these patent-protection deals Microsoft has made with Linux vendors really had nothing to do with patents and IP at all — the real point of the deals was to get the vendors' cooperation for increasing open source interoperability with OOXML, thus keeping Microsoft's precious office suite market safe from incursion from OpenOffice.org.

If this is indeed the case, expect to hear much hollering from OpenOffice.org's primary commercial benefactor, Sun Microsystems, very soon. If states are allowed to fall for the notion that OOXML is actually open, OpenOffice.org's potential growth onto government desktops could be cut short.

Curiously, another possibility as to why Microsoft is so pro-IP these days has to do with something that happened long before Linux was a gleam in Linus Torvalds' eye — something from the middle era of the history of Unix, when Microsoft itself had its own Unix, and proceeded to charge royalties for it in perpetuity. It popped up this weekend, then promptly got lost in the GPLv3 announcements.

The Unix I am talking about is Xenix, derived from Version 7 Unix, which Microsoft licensed from AT&T in 1979. Xenix, launched a year later, ran on 16-bit processors and was sold to various OEM vendors like Altos, Tandy and the Santa Cruz Operation ... the company that would later become The SCO Group. Because of its OEM placement in the market, some have indicated that Xenix was one of the most distributed flavors of Unix, ever.

In 1987, with Microsoft getting all hot and heavy for OS/2 with IBM, Microsoft sold Xenix to SCO — but with a catch revealed just this week. While it's been long-known Microsoft ended up owning 25 percent of SCO as a result of the deal, Microsoft also was granted royalties for all future sales of Xenix. This was revealed, naturally, on the Groklaw Web site, which will likely be one of the most complete archives of the early days of Unix by the time the various SCO Group trials are over.

The document, submitted in The SCO Group vs. Novell trial, was a letter "that an outside attorney for Santa Cruz sent to the Department of Justice in 1996, complaining about a 1987 agreement between AT&T and Microsoft which [SCO] says it had inherited, with terms such that [SCO] found itself contractually compelled to include outdated Microsoft Xenix code forevermore in all its own Unix products and to pay royalties on the undesired and irrelevant code."

This is very interesting because if it were ever to be shown that some of The SCO Group's code actually ended up in Linux (which at this point isn't very likely), then, one could argue that some of The SCO Group's inherited Unix code from SCO's Xenix (and Microsoft's Xenix before that) is also in Linux.

Groklaw is not the definitive site on what might have happened with all of these business arrangements and code sharing, any more than any other site. It does, however, do a pretty good job of ferreting out some of the arcane stuff from the deep IT past. Therefore, dismissing this idea isn't that easy. Plus, it pulls together some things that didn't make much sense to me at the time.

Back in spring 2004, a think tank known as the Alexis de Tocqueville Institution was touting a new book, "Samizdat: And Other Issues Regarding the 'Source' of Open Source Code," which, among other claims, said there was no way Torvalds could have come up with the Linux kernel on his own. In that case, the book stated, he must have gotten the code from the Minix operating system, developed by Andrew Tannenbaum.

At the time, the book's author, Ken Brown, came under heavy criticism for these assertions, since even Tannenbaum stood by Torvalds and denounced such code-borrowing. Along with several members of the Linux community, I got into a verbal tussle with Brown on the old Internet broadcast "The Linux Show."

Later that summer, it was revealed that one possible source of Brown's misinformation was the History of Unix chart maintained by Eric Levenez. Actually, not the real chart but one hosted on The SCO Group's Web site. In The SCO Group's version, there was a line of connection between Xenix, Sinix, Minix and, finally, Linux. (It would be Groklaw again that pointed this out, by the way.) If Brown got his info from the wrong chart, then no wonder he thought Linux was derived from Minix. After all, if it's on the Internet, it must be true.

It is not clear if the current incarnation of SCO, The SCO Group, is still entitled to pay Xenix royalties. But Microsoft still does a lot of business with The SCO Group, and if it could ever find a direct line between Xenix and Linux, it would make for a very powerful weapon to kill off Linux once and for all.

The incentive is powerful enough to speculate on the possibility that such a direct ancestral line could be created, rather than discovered.

Brian Proffitt is managing editor of JupiterWeb's Linux/Open Source channel, which includes Linux Today, LinuxPlanet, and AllLinuxDevices.