[sf-lug] Potential GPL violation, with the potential intent to victimize our Elderly loved ones

[Jesse Zbikowski]

I guess we are all clear that the particular case of Eldy does not
redistribute GPL code improperly. It's interesting to take the
opportunity to explore what rights copyleft is meant to protect, and
whether the GPL does a good job of it.

On Fri, Nov 13, 2009 at 4:59 PM, Rick Moen <rick at linuxmafia.com> wrote:
> You might find it of interest that the criterion of "has knowledge of
> and plays with fundamental internal Linux behaviour" doesn't hold water.
> (Yes, I know that's what Torvalds and some others talk about.)  It's
> apparent _why_ people think that:  The theory is that coders probably
> could not engineer that degree of close cooperation without copying code

I see two justifications for this "fundamental internals" criterion.
The first is just as you say: that people presume some code copying
must be involved, which is in fact generally the case. I agree that
the type of code copying is not necessarily copyright-protected so
that the module developer may not need a license at all (especially if
he were to re-write the needed header files).

The second, which I mostly see Linus arguing for, is that the binary
module is presumptively a derived work because it is only meaningful
the context of the original work it is derived from.  That is, it is
not a complete expression of an idea, but a modification (or
elaboration) of the expressions found in the original work. I'm not
sure how legally solid this argument is.  It would seem to imply that
if I make a modification to a vacuum cleaner design, and that original
design is under copyright, I cannot sell my modification without
permission.

To put a finer point on it: if the vacuum cleaner design had a clearly
demarcated interface for attachments, I could clearly distribute a
standalone attachment useful on any vacuum cleaner supporting the same
interface.  But if I am simply "tweaking" the internals of the vacuum
cleaner rather than "adding" to it, to the extent that my tweak is not
separable from the original, the tweaked design is a derived work.
Linus would say in the OLD days, the kernel module interface was so
clear and limited that module code might be considered separate from
kernel code; nowadays the interface exposes essentially EVERYTHING, so
module code is kernel code is GPL.