2006-02-15 (水) の 00:25 -0600 に Les Mikesell さんは書きました: > On Tue, 2006-02-14 at 23:50, Joel Rees wrote: > > > The only problem with the GPL is the misinterpretations that circulate. > > There are misinterpretations, but they aren't as much of a problem > as the FSF's interpretation. ?? > > The GPL allows dynamic linking to incompatibly licensed software, as > > long as the separation is clear, and the basic requirement is that the > > API should be standardized enough that there should be at least a > > theoretical possibility of swapping what is linked out for something > > else. > > It boils down to what might possibly be considered a derived work > under copyright law. partly > The FSF has taken the position that if > a GPL'd library is unique, then anything that uses that > library is a derived work and thus subject to the GPL > restrictions even if it is distributed separately from > the library. I think you are overstating the case. > In the past, an author that wanted to > give a work away freely without the GPL restrictions was > forced by the FSF to rewrite a library (badly, it was > never really used) with all the corresponding functions > instead of just permitting users to link their own > GPL'ed library obtained separately. Are you speaking of an actual case, or an actual FAQ, or are you speaking of somebody's panicked interpretation after having used GPLed source without having read the license? Anything which does not incorporate is not derived. Static linking actually brings in the library code, so it is derived. Stallman describes his take on dynamic linking, it's there on either fsf.org or gnu.org for anyone to read. Publicly stated. Anyone worried about it can read that and see what they think for themselves. Until there are court cases, of course, it's just like any other license -- we don't know what a court might decide, and we put ourselves somewhat at the mercy of the license holders when we use their property. So what? (I personally prefer to depend on the mercy of someone who publishes under GPL more than of someone who publishes under Microsoft's EULA, for example.) > > If this weren't the case, no proprietary software could run on > > Linux, period. > > No, the reason this is permitted is that other implementations > of the standard system libraries exist to prove that the > programs are not derived from GPL code. There is, however, > the potential claim to be made against anything that uses > a uniquely-linux, uniquely GPL'd function. Potential claim? Oh, yeah, a certain level of paranoia is good, but it sounds to me as if you are basing your interpretations on the assumption that anyone who publishes under the GPL is likely to be hiding the intent to pull an "embrace and extend" maneuver. To which, the answer is to fork. (Sure, it takes work.) > > If the above were not the case, Mac OS X would not legally be able to > > run any propietary software. It wouldn't even be legal for Apple to > > compile iTunes on it or for it, considering Apple's compiler is the GCC. > > Open Office is another example, also Cygwin, ... > > Once again, the C library is a standard C library. And in the > OSX case any unique kernel functions will not be covered by > the GPL. > > > Stallman does have the goal of seeing a world where the LGPL is no > > longer needed. Won't happen this side of the Millenial Reign, but it's a > > nice goal. > > It's not at all nice for the people forced to use only one > or the other. You are always free to ignore the GPLed software, which was the way it was before the software was published under the GPL. Freedom has its limits. Wind friction, gravity, tire friction, ball-bearing friction, all these things make it hard to pump your bike. But can you pump in a vacuum? If the bearings have absolutely zero friction, could you keep them in the axle? If there were no gravity, how do you keep your tires on the road? The GPL specifies certain limits. I guess you find them uncomfortable. Over in OpenBSD land, there is more "freedom", but Theo enforces a compatible license on anything in the project trees. Different kinds of limits. FreeBSD has its own limits, although they might not be obvious (until you go to update the system and packages). You pick your freedoms, and then you work within them, or you don't work at all. > > Still, he has left plenty of room for people who don't > > understand that money and value are not equivalent to use GPLed > > software, as long as they don't try to abuse the largess of the > > programmers that wrote it. > > Fortunately there are people who don't feel abused when > people use their code in ways they didn't plan. I consider > the perl license to be one of the nicest examples. The > dual-license option keeps it from being trapped by the > GPL's viral nature yet still allows it to be combined with > GPL code. If the GPL bothers you, why do you use Linux instead of the BSDs? I recommend them on their own merits, and, with a little work, you can get almost anything that runs on Linux to run on those. If you find Linux's licensing so distasteful, that is. I personally don't find the GPL distasteful. I'm sure I might be frustrated by licensing at times, but that's the one of the prices we pay here. (I find my own time and financial limits much more frustrating.)
