In comp.os.linux.advocacy, spike1@xxxxxxxxxxxxx
<spike1@xxxxxxxxxxxxx>
wrote
on Thu, 13 Jul 2006 08:39:32 GMT
<uadho3-uui.ln1@xxxxxxxxxxxxxxxxxxxx>:
> Roy Schestowitz <newsgroups@xxxxxxxxxxxxxxx> did eloquently scribble:
>>>> XGL is awesome! But what is the purpose of your question?
>>>
>>> I think he wants a cd bootable live distro that has nforce built in.
>
>> Whichever distro it may be, this will probably be a GPL violation.
>
> How would it violate the gpl?
> The gpl spefically distinguishes between a piece of software and a
> distribution of software components. The kernel might be gpl, but it doesn't
> stop suse from shipping with opera, for example.
AIUI, the kernel is shipping with a modified GPL.
The modifications allow for programs to execute without
the entire system being considered a "derived work".
Unfortunately, /usr/src/linux/Documentation doesn't appear
to contain a EULA or License file of any sort. There is a
mandatory.txt but it's describing a technical subfeature,
namely, mandatory file locking using flock()/fcntl().
No COPYING or COPYING.txt, either.
A quick Google coughed up two entries.
The Wikipedia entry suggests it's developed under a
standard GPL, which is admittedly possible.
http://en.wikipedia.org/wiki/Linux_kernel
If it *is* shipping without a license file, or explanatory
text on how to get it, it's in technical violation, though.
http://www.kernel.org/pub/linux/kernel/COPYING
describes a statement by Linus explaining the
exemption/discrepancy/non-coverage, followed by what
appears to be a stock GPL. I've not studied the text
of this copy of the GPL in detail, though.
There is a file /usr/src/linux/COPYING, which is apparently
where Linus and crew decided to put it. Phew...it's not
in violation after all; the license just got stuck in an
unexpected (by me) spot. :-)
But it gets weirder: the file in my kernel (a stock 2.6.14)
has an additional paragraph:
Also note that the only valid version of the GPL as far as the kernel
is concerned is _this_ particular version of the license (ie v2, not
v2.2 or v3.x or whatever), unless explicitly otherwise stated.
One might call this "defensive documentation", I suppose. :-)
(The extreme pedant in me, however, is asking how the
kernel might solve tough philosophical problems, find a
mate, and reproduce. It's a gray area.)
Of course this modification has no bearing at all on the
problem anyway; the nvidia/nforce modules are modules,
not executables, and AFAIK do not use kernel services by
"normal system calls", though I'd have to look to see how
modules get at things like sockets (necessary for NFS work)
and device I/O (I do know about bread(), which stands for
buffer read, not a baked goods product :-) ; there's a
corresponding bwrite() as well).
It would appear to me that shipping nvidia/nforce as part
of a Linux-based system is a "derived work", in the sense
that they've slapped something onto a stock distro. Whether
the Linux kernel license or a distro license would therefore
be in effect is an interesting jurisdictional matter.
I frankly don't know.
--
#191, ewill3@xxxxxxxxxxxxx
Windows Vista. Because it's time to refresh your hardware. Trust us.
|
|
