2008/10/13 Neil Williams <[EMAIL PROTECTED]>: > On Sun, 2008-10-12 at 18:06 +0200, Roberto A. Foglietta wrote: >> >> > You have to comply with GPL v2. Which is not difficult and costs >> >> > nothing. >> >> >> >> In case you want deliver your specific proprietary command line >> >> executable and you would like to keep its size very small then you can >> >> compile it linking against busybox library. Remember that GPL allow >> >> only dynamic linking, static should enforce GPL redistributions terms >> >> and make your application bigger. >> > >> > Eh? The GPL does not allow dynamic linking against non-free code. Are >> > you thinking of the LGPL? >> > >> >> Thanks very much for your promptly correction. I forgot to say that >> the point of view I exposed is NOT the strictly/cautionary one. > > It is not even logically coherent, IMNSHO. This isn't just about how the > library is used at runtime. Using a public shared library involves > building against that shared library and this means *including* the > source code of that library directly into the program. > #include <qpl-header.h> > does exactly that - the preprocessor physically includes the entire > header file and all header files that it references into the object > files being compiled. The header file is copyrighted and licenced under > the GPL. The licence forbids the copying of the file and it's contents > except under the GPL. If you grep the resulting binary (even after using > strip), you will find that elements of that file are retained in the > executable - the (copyrighted) names of the symbols that are defined in > the GPL library. If that executable is not licenced under the GPL, that > contravenes the GPL. Simple. At the compilation stage, the GPL'd code > has become a part of the executable at a *physical* level that can be > easily identified later. > > The question is whether the non-free code can exist without the GPL > code. If it cannot be built without the GPL code then the GPL code is, > by definition, included in the executable - part of the executable, the > executable and the library are one program. The mere fact that the > object files are split into different files for convenience does not > detract from the fact that the symbols from the library are physically > part of the executable and therefore that the executable is derived from > the library - namely the header files and the definition of the > copyrighted symbols contained within. The linking comes later, the > problem is the inclusion of the copyrighted symbols and the licence > under which those symbols can be used. That is why I said that if the > library shows up in 'objdump -p' as NEEDED then the executable must be > released under the GPL. > >> It was >> a HUGE missing I recognize it, please do not flame for this. In facts >> there are people who consider acceptable dynamic linking against GPLv2 >> (or previous) libraries. > > I've read the refs you mention, I do not think that any are logically > coherent. It is wrong to advise people that they can link when logic > dictates that they cannot. > >> You are right in say that GNU indicates that >> dynamic linking is not allowed but unfortunately they could be wrong, >> at least in writing correctly the GPLv2 many years ago. > > Until that is tested in court, it is mere speculation. The people > enforcing the licence in court will be the FSF so it's not exactly > sensible to dismiss them so lightly. >
Unfortunately as Rob said much more people are going to brought in front a court in these days because GPL. I feel that as long as the courts does not set a coherent and long record of judgements the debate continues. I think it is perfectly sane and correct FSF would try to enforce their view of GPL. Unfortunately I think is true what Novell/SuSE has written in their site: in future we will see a growing of dynamic linking against GPL software APIs (some smart, some malicious and some stupid). Unfortunately Novell are not such insane because being sued in U.S. is very costly while in E.U. is pretty cheap, compared. The FSF has a moral point of view when others have an economic point of view: the tipping point in which people agree with FSF is when they realize that behaving in a correct moral way results an economic advantage too but it depends on the costs of the risk of being sued. Unfortunately Novell seems to know this and it is a fact as real as the FSF enforcing willing. I was in good faith when I said it is considered fair the dynamic linking with GPL code. This could be because a) I have been confused GPL with LGPL or b) I have been exposed to a lot of speeches and material in which dynamic linking is considered fair. In case of b) I am a victim of FUD. However I feel that someone is going to collect, at least in EU, a critical mass of economic interest to brought in courts. Such critical mass would turn up the verdict in favour of Novell's point of view in the same way U.S. citizen has going to pay 700 billion of dollars because too many banks fail. Moral is not a fixed thing but evolves in the same way consensus does. When difficult technical questions are brought in front of a courts judge would relay on experts and experts relay on their credibility and credibility evolves in the same way consensus does. Driving the consensus correctly could achieve results that logic does not. I have understood your critics and I hope I was just confusing the LGPL with GPL. I happy to inform you that I am pretty sure that all source code in which I have been messed was carrying a GPL or LGPL copyright statements in the head of the files otherwise I have noticed it on my agenda. So, in both a) or b) cases nothing bad would happen because of me. Sorry, -- /roberto _______________________________________________ busybox mailing list [email protected] http://busybox.net/cgi-bin/mailman/listinfo/busybox
