SCO has sent out a letter to high-profile Linux users specifying some of the files they say infringe on their IP rights. It’s funny stuff. Why? Because while, to a duplicate-finding computer program or your average lawyer, it looks like there’s infringement, it’s perfectly obvious to anyone who understands both copyright law and the functions of these files that (1) they aren’t necessarily copied from UNIX System V, and (2) even if they were, there is no copyrightable matter contained in these files.
Let’s take, for example, errno.h. This is simply a list of error numbers and their meanings. Among the “60 files” revealed, 17 of them are this file. There are multiple copies of it in the Linux source tree, one for each major platform to which Linux has been ported. That’s why you see the same file name, preceded by things like asm-i386, asm-ppc, and asm-arm — those are the error number lists for Intel x86, Motorola PowerPC (used in Macs), and ARM processors (used in devices like Palm handhelds). They’ve really only pointed to a few files here.
Any programmer who’s coded in a UNIX-like environment recognizes these as the standard error numbers and descriptions. They’re so standard, in fact, that they’re part of the free, open POSIX / Standard UNIX specification. The error names and their descriptions are posted for free use by The Open Group here. All that’s missing are the numbers themselves, and the numbering alone is insufficiently original to warrant copyright protection.
I’ve taken a look at some of the other files, and there seems to be something similar going on in them as well — these files are identical to files in System V UNIX because there’s really only one good way to implement these files, and they’re required by the POSIX specification. Idea and expression merge, so there’s no copyright issue. SCO is blowing more smoke.
I seem to recall somebody making a joke that the “hundreds of copied files” would just be dozens of identical copies of uncopyrightable header files. These jokes just keep on turning out to be true . . .
UPDATE: Linus wrote this refutation of SCO’s claims of infringement. Turns out he didn’t use the same error numbers, he got much of the content of those files out of a standards book published by Intel, and the few more complex routines that are the same are textbook idea/expression merger issues.