Re: I think Apple owns Unix now anyway @lars
I would actually dispute that UNIX(tm) has ever been Open, as we would think Linux or other GPL code is.
Yes, UNIX code source code has been available, but only under license. Versions (editions) 1-6 were available to academic users under a very permissive license, but one that prevented commercial use. At the time, Bell Labs/AT&T were prevented by a US anti-monopoly judgement from supplying commercial computers, and this included Operating Systems. At this time, there was a thriving pre-Open Source group of academic users who dabbled in the code, and shared their work with others. This was a really exciting time (I was there), and you often found 400' 1/2" tape reels being sent around (it was pre-networks) various Universities.
Version 7 tightened this up to prevent the source from being used as a teaching example. Version 7 and earlier code has, since 2002, been published under GPLv2, granted by Caldera (Horray!). This is now "Open", but I don't know of anybody who is shipping a commercial V7 implementation (although a free x86 port is freely available from a South African company called Nordier Associates).
Commercial use of UNIX post Version 7, from PWB to UnixWare was under a commercial license that did not contain any right to the source code. The same was true for all other-vendor UNIX systems. Source licenses were available, but under their own strict licensing conditions, and at a high cost (and often required the licensee to have an AT&T source licence as well!).
BSD code prior to BSD/Lite required the user to have an AT&T version 7 (or later) license. BSD/Lite or later does not contain any AT&T code (or at least nothing that AT&T were prepared to contest), so is available under the BSD license, but as I have stated before, cannot legally call itself UNIX.
Having got that out of the way, why was UNIX used as the basis for Open Systems?
Well, UNIX was always easy to port. This meant that there were several vendors (piggy-backing on various academic ports, like SUN and DEC) who could sell UNIX systems, meaning that application writers have something approaching a common base to target their code, although differences had to be worked around. This was unique. There was no other large-system operating system around at the time that had this.
It became apparent that if there could be a standardised subset of UNIX (commands, APIs, libraries) that all vendors would support, then this could mean that application writers could possibly entertain a "write-once, compile once per vendor UNIX, and sell" strategy. This was first championed by AT&T (who by this time were allowed to sell computers and operating systems) with the System V Interface Definition (SVID), which was adopted by IEEE, with minor changes, as the various POSIX 1003 standards.
These standards are what gave UNIX the "Open" label. Anybody could write an OS that met these standards, whether based on genetic UNIX code or not. This has resulted in numerous interesting products and projects, one of which is GNU/Linux (POSIX compliant, but not any later UNIX standard), and includes such things as QNX, BeOS and z/OS, which can be regarded as UNIX or UNIX-like, some of which are truly open. Not all of these can be called UNIX, however.
I agree about the Linux kernel. The reason why this has remained as a single kernel is because Linus keeps an iron hand on the kernel source tree and official release numbers. It is perfectly possible for someone to take this tree, and modify it (and it has been done by several people including IBM and Google) under the GPL, but they can't get their modifications back in to the main tree without Linus' agreement. They could maintain their own version, however, as long as they abide by the GPL. AFAIK, they can even still call it Linux.