pcg@aber-cs.UUCP (Piercarlo Grandi) (06/07/89)
In article <9287@csli.Stanford.EDU> carl@csli.stanford.edu (Carl Schaefer) writes: In article <4818@uoregon.uoregon.edu> markv@tillamook.UUCP (Mark VandeWettering) writes: > It sure would be nice to replace those copyleft notices which > add a good 8K to every file in GNU software with something like > > "Use this code. Improve this code. Tell me about it." This is precisely the point. The copyleft allows the first of these, encourages the second, but requires the third. Actually, it does not require the third. Nobody is under any obligation to distribute material that includes parts of FSF material. You can develop all kinds of proprietary material that includes FSF's, and keep them to yourself, and tell nobody. The obligation you have is that *if* (and only if) you distribute to third parties executables that include any nontrivial amount of copylefted FSF material, you must be ready to supply the sources for such executable (yours and the FSF's) at the conditions detailed in the GPL. There is emphatically no obligation for you to distribute your material to all comers, or to let others know it exists (while such obligation exists for Academic Source licensees of Un*x, if you decide to distribute sw based on licenced material). There are two further subtle points: [1] as far as I can see, the GPL conditions apply to the sources, but *not* to the executables (which is mostly a moot point though); [2] you have at least two ways (actually two variations of the same idea) of "cheating", without violating the terms of the GPL, but they are cumbersome, so much so that it is probably pointless to use them in a commercial environment. [the idea is that you can supply proprietary extensions, patches, modifications, libraries to a customer that then proceeds to incorporate or link them into their copy of FSF software. Since you supply material that is neither a proper derivative nor incorporates any FSF material, you escape the terms of the GPL.] Last but not least the compilers: as RMS has stated many times (most recently here), the output of the compilers is not and cannot be copyrighted, and their fundametal RUNTIME library and header have been put by the FSF in the public domain, they are not copyrighted. [note that C (and almost as same is true for C++) does *not*, by definition of the language (at least until Ansi C defined an "hosted" implementation), include os interfacing, io, string handling, memory allocation, etc... the runtime library proper therefore only includes code to support language primitives, e.g. prologue and epilogue, variable argument lists, long arithmetic for some machines, etc...] The FSF does supply also a copylefted STANDARD (i.e. what is described in chapters 7 and 8 of K&R) library (and many others, like Doug Lea's suite for C++), though, for your convenience. If you include *this* sw with your programs (whether you compile them or the libraries with an FSF compiler or not is immaterial) then the GPL applies. If you want to avoid this, use another "standard library"; there are many free ones, or you can use one by a commercial supplier (e.g. the one often included with your os), maybe one like AT&T, which lets you use their libraries for your products without royalties. The FSF compilers take care to generate code that can be linked with that compiled by other compilers, (even if sometimes though the default linkage convetion is different), and that can be used with precompiled libraries by other parties. [if I have misunderstood the intended meaning of the FSF conditions, I hope some of them will correct me]. -- Piercarlo "Peter" Grandi | ARPA: pcg%cs.aber.ac.uk@nsfnet-relay.ac.uk Dept of CS, UCW Aberystwyth | UUCP: ...!mcvax!ukc!aber-cs!pcg Penglais, Aberystwyth SY23 3BZ, UK | INET: pcg@cs.aber.ac.uk