>> Russell Cattelan <cattelan@xxxxxxxxxxx> writes:
> What you are saying all of the non GPL'ed linux kernel modules are in
> violation of the GPL?!
One of the main copyright holders has said that it is ok to "link" the
GPLed code with code under a license that conflicts with the GPL.
Other copyright holders have not said they are against this, but any of
them can say so. With the Linux kernel everyone is acting in good
faith.
> Guess somebody better stick some lawyers on Nvidia.
As an individual, you can't distribute NVIDIA's drivers in any form
(unless they have changed the license drastically lately, but I'm sure
they haven't). Once you get NVIDIA's permission to do so, you can
distribute the drivers. NVIDIA provides source to some parts of the
kernel module because it's easier for them that way, not because they
have to. As an individual, you can you do whatever you want on the
privacy of your own computer. If people bothered to read the GPL (or
any other license for that matter) they would notice it's a
*distribution* license, not a use license. You can modify a GPLed
program as much as you want, you can use pieces of it in your own code,
you can link it with whatever you want, and it's all ok as long as it
doesn't leave your computer (and *please* let's not enter a discussion
regarding what "leaving the computer" means, this is off topic enough
already)
> Ohh and explain this problem to Linus also.
Linus has given his own interpretation of what the user can and can't
do (distribution wise). He is free to do so as copyright holder (in
the same way SGI is free to "clarify" what the GPL means according to
their lawyers -- everyone here knows that SGI has provided an
interpretation regarding a couple of points of the GPL, right?). This
doesn't mean he is right wrt what the GPL allows you to do or not.
> > Sun Solaris was just an example. The number of "threats" to SGI's
> > IP would be a lot greater than you think if they released XFS as BSD
> > licensed.
>
> Ohh?! like how...
Is that a serious question? By providing the XFS code under a license
not as loose as the BSD one, SGI is ensuring they keep some degree of
control regarding what the competition can and can't do. HP can take
the Linux kernel, include the XFS patches and distribute it with the
shiny Itanium servers, and it's all ok (funny, this sounds a lot like
what SGI wants to do, too). Along the same line of thought, Sun can
take the XFS code, patch it to get it to run with Solaris (probably
modifying Solaris a bit at the same time) and distribute that (not as
part of the Solaris kernel, but as a loadable module). This is
probably not ok according to the GPL (this is not the same as a GPLed
program linked with the vendor's libc, this is linking GPLed code
inside kernel space) and it probably requires permission from SGI to
distribute. At any rate, iff SGI is ok with this, and say, Sun finds a
way to make XFS run ten times as fast, they *have* to give the source
to the people that get the binary module, and SGI can get the
modifications back.
Honestly, it can't get simpler than that. In the context of large
projects with a large interested audience, the BSD license is business
wonderland for the people *getting* the code, not for the people
writing it. I would have thought that people doing consulting work
would understand this without further explanation.
But yes, "threats" is a bit off the scale.
--
Marcelo
|