On Wed, Mar 05, 2003 at 12:58:17PM +0200, Eli Billauer wrote: > You don't close the source of the kernel. You close one small crucial > component of it by rewriting it. This means, that every time a new > kernel is released, you take its source, make a patch, compile, and > remove your own little component's source. And distribute.
You need to look deeper into what "derived work" means in the GPL. In this specific case, it means that unless you cripple the kernel (for example, by taking the scheduler and moving it to user space completely), you just can't do it. > Does this violate the GPL? I mean, everything that came free is released > free, and the kernel comes along withonly one small binary file. We've > *added* something, which isn't free... Derived work is not only about modifying, it's also about removing and adding. IANAL, etc. > But we don't have to go as far as the kernel: You can release a link > library in closed form. Which will do what? and how long will it take until you'll have an open source counterpart? > Assume that Microsoft released some link library in closed form. Then > they released Office for this special Linux edition, and made sure they > use this special library. Now they can sell Linux like any other > operating system. No, they are not selling the OS, they are only selling > their own link library. Which means that they can take any open distro, > put their library on, and resell it, with copyright restricted on this > tiny library only. All true, so far. So what? they are not "subverting" Linux, they just compete under the rules of the game like everyone else (in this case!) and good luck to them. Your original argument was completely different, remember? > Microsoft has the powers to convince other software vendors to rely on > their link library as well, so you end up with one Linux edition, which > can run common applications, and the open one, which can't. Again, so what? > They won't, of course, be able to control the open version of Linux, but > they will control the *standard* edition, which will soon enough divert > more and more from the open one. They've tried to do it in everything they've done so far, and maybe, just maybe, they're failing (Cue the next Office using "open standards", cue Arie Scoop talking about "open standards". Who cares if their version is the standard one, as long as it can interoperate with whatever *I* want to run? Interesting argument, but you haven't convinced me yet :-) -- Muli Ben-Yehuda http://www.mulix.org
pgp00000.pgp
Description: PGP signature
