Re: [Gnu-arch-users] [OT] SCO vs. the world

[Miles Bader]

> 1) Sequent had license to unix.   They could distribute modified
>    versions only if their modifications were, in essense, contributed
>    back to unix and controlled by the same license.
> 
> 2) They wrote a lot of code that fell under those terms.
> 
> 3) IBM bought Sequent.
> 
> 4) Lots of that code showed up in the Linux kernel, in violation of
>    the terms of the unix license.

I think it's hard to judge these issues unless you have actual information,
as the details (of both the source involved and the licenses) matter _a lot_
-- and of course none of us has any.

> I tend towards believing the large LOC counts.

I don't:

 (1) The functionality involved is not a very large portion of linux, and
     code that goes into central areas of linux usually gets massively
     scrutinized and mutated before it actually goes in.

 (2) This sort of code is interwined with many nit-picky details of the
     kernel implementation, and linux is unlikely to have much in common with
     SCO, other than when viewed from the 10,000 foot level.  This means that
     code copied from another system typically _can't_ be used as-is, and
     making it fit involves pretty serious restructuring.  The most likely
     thing to survive this is probably random trivial helper functions and
     the like (though random trivial code snippets and comments might
     also make the cut).

 (3) To the best of my knowledge, a lot of stuff related to this area has
     been changed a great deal since 2.4.x

The above comments pertain to the theorized SMP/NUMA breaches; I've also seen
it suggested that IBM's JFS filesystem is involved, which, being a more
self-contained body of code, may be more suspect -- but also far easier to
excise.

> IANAL but to my ears, it sounds like a serious breach and it does sound
> like a decent percentage of kernel code isn't legitimately GPL'ed,

Naw, the possible problem areas are a very, very, tiny proportion of linux.

> particularly in economically significant areas of functionality.

_If_ the NUMA support turns out to be a problem, it might be annoying --
I gather than linux's support for such systems is a big brownie point with
large powerful corporations, though not particularly significant for the
general user base -- but I think that's all it will be.  IBM may owe SCO some
(or a lot) of money, and the code involved will have to be rewritten, but
SCO's really significant claim, that they somehow own basically everything,
is clearly silly.

-Miles
-- 
"I distrust a research person who is always obviously busy on a task."
   --Robert Frosch, VP, GM Research