On Thu, 2008-10-23 at 21:13 +0200, Vincent Danjean wrote: > Ben Hutchings wrote: > > On Wed, 2008-10-22 at 11:38 +0200, Bastian Blank wrote: > >> The iwl4965 firmware changed 2 times incompatible since the driver > >> exists. > > > > That makes me wonder just how separate the driver and firmware are. If > > they are tightly coupled then the firmware may become subject to the GPL > > as well. > > Firmware and driver do not run on the same CPU. There is no 'linkage' > between them. With a client/server application, a GPL client does not > enforce the server to be GPL, even if client and server are tightly > coupled.
That is not true. It simply depends on whether they are one program or not, which is a human-level concept, and not a technical one. There is no "magic boundary" at which the GPL would neve cross. For example, if you were to split GCC into two executables, one which parsed and generated intermediate code, and another which did optimization and codegen, the result would still be the one GCC, covered by the GPL. And this is true even if you then write your own version of the first part, implementing your seekrit proprietary language: the GPL on the back end would require that the *whole program* be distributed under the GPL, any separation into different executables notwithstanding. There is nothing in the GPL about "running on the same CPU" or "client/server" exceptions. If you use GPLd code, then the *whole program* (whatever that is, it is a human-level concept requiring understanding and not rote following of rigid rules) must be distributable under terms no more restrictive than the GPL itself. Thomas -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
