On Mon, 2006-06-12 at 17:51 -0500, Jan Depner wrote: > On Sun, 2006-06-11 at 13:34 -0400, Lee Revell wrote: > > On Sun, 2006-06-11 at 01:29 -0500, Jan Depner wrote: > > > That's the consensus among kernel people. Not lawyers. Take it to > > > court first. It has zero chance in that venue. > > > > Can you cite any precedents? > > > > The kernel people I am referring to have talked to their lawyers about > > it and the consensus is that a driver is a derived work of the OS that > > it is developed for. > > > > If the driver is part of the kernel then that is true. If it is a > module then it isn't. The whole thing is moot though until someone > takes it to court. I seriously doubt that anyone will be willing to try > to win that battle when they face the possibility of losing and being > counter sued for court and legal costs. > Built-in vs. module is irrelevant - modules are linked into the kernel at runtime. See: http://kerneltrap.org/node/4674 http://lwn.net/2001/1025/a/module-license.php3 http://lwn.net/Articles/154602/ > > > Look at some driver source some day - it's basically impossible to write > > a driver without using any kernel APIs - driver model, spinlocks, etc. > > > > My understanding of the NVIDIA driver is that it uses an open module > to work with a closed module (BLOB). What you have been saying is that > the intent of the vendor to only develop the closed module for Linux > makes it a violation of the GPL and this is obviously not true. There > is no infringement in the closed module. They're not using any GPL'ed > code. This is why we have binary modules now. At any rate, as I said > above, until someone wants to take it to court it doesn't matter. > NVIDIA will continue to make closed drivers and so will other vendors. > As I've stated many times I am not talking about the nvidia driver. That is legal because the binary part cannot be a derived work as it was developed for another OS. Lee