> If the problem's what I think it is (which it might not be!), it's that,
> with CTM & NO base source files, you could get a file (or large pieces)
> w/o the original copyright (since it's not changing, there aren't diffs on
> it). Lookaing at sys/kern/init_main.c, in the BSD copyright, item 1 is:

Well, more importantly, if you had a source file with a different
copyright, and you read just the diffs, you could use them to apply
fixes from the main line to your copy of the file, and theoretically
not be covered by the copyright.   Having just spent a bunch of time
finding incompatibilities between NetBSD/Alpha and the MILO PALcode, I
can say with some authority that the majority of time is spent finding
and figuring out these incompatibilities, not writing the code to work
around them.   So these diffs do generally represent *significant*
work.   Ensuring that they are covered by the license, be it BSD or
something else, is worth some effort.

However, my suspicion is that looking at the diffs to see what changed
could be construed as fair use in any case, so I'm not sure what this
really means in a legal sense - it may be that even if every diff
contained the license for the file to which it applied, it wouldn't
make any legal difference.   I'm not a lawyer, so I don't know the
precedents in this case, if there are any that apply (I doubt it).

			       _MelloN_