But ljl has posted as neat an analogy for the fiasco as I've heard over on Iconoclast Blast. While her summary is humorous, and true, as far as it goes, I think most folks are missing the meat of the matter: derivative works. This is the issue upon which AT&T's Unix Systems Laboratory (USL) lawsuit against UC Berkeley and BSDI turned. The original Unix license provided that "derivative works" were covered under the original license, and could not be revealed to third parties. In that lawsuit, it was shown that Berkeley Unix was comprised of mostly new code, and that Bell Labs had copied a lot of this code back into System V, removing the university's copyright notice as they did so. The case was never adjudicated, but was settled by Novell soon after it acquired USL from AT&T.
Many point to that earlier lawsuit to bash SCO for filing against IBM. But the USL vs BSDI case was never finished, as noted above. And IBM had a different Unix License than UC Berkeley did. It now appears that there was a side agreement between AT&T and IBM that acknowledged IBM's rights in Unix software that it developed. But the situation is complicated. Much of the code SCO is apparently complaining about (they aren't actually saying which code it was, exactly) was developed by Sequent, which was bought by IBM. Sequent had its own Unix license, the terms of which will bear on this lawsuit. The waters are muddy enough that it is possible that SCO could prevail in its lawsuit against IBM. IANL, and even if I were, I wouldn't be able to give a good assessment of the various side's chances without access to a lot of information, most of which won't be available until discovery and pre-trial motions are complete.
Nonetheless, ljl's analogy of SCO being the bad sport who wants to take the ball away, even though he doesn't own it seems apt to me. SCO's attempt to extort license fees out of Linux users seems to have very little basis in law, even if SCO prevails against IBM. (IANAL, already!) And IBM must feel pretty good about its case if it is willing to risk AIX, as it appears to be. But if I were Linus Torvalds, I might want to start looking into just starting projects to replace NUMA and RCU, two of the three technologies contributed to the Linux kernel by IBM. Actually completing such projects would be very time consuming. But why not tool up, just in case?
Posted by hbo at August 2, 2003 09:01 PM