G'day Deepayan, >>>>> "DS" == Deepayan Sarkar <[EMAIL PROTECTED]> writes:
DS> On 5/21/06, Berwin A Turlach <[EMAIL PROTECTED]> wrote: DS> A user can never violate the GPL. The GPL does not govern use, DS> it governs distribution. Specifically, >> As I said, I stopped reading gnu.misc.discuss long time ago, >> but if I remember correctly sometimes in the (early?) '90s the >> following case was discussed. >> >> A company made a binary module available for download and gave >> the instructions "go to the FSF site (or a mirror), download >> version X.Y.Z of program U and compile it with these options, >> then link our module and start the program, now you can use >> these features of ours and are in business". (Remember, these >> were the days when most "free" software was only available in >> .tar.gz form, people were used to compile their own software >> and slackware was the dominant (only?) linux distribution, no >> .rpm or .deb files.) >> >> Note also that they did not distribute any GPL code, they said >> go and get it. As far as I remember, they were told by the FSF >> that they cannot do this and had to stop. And, IIRC, the >> argument was that whether they did the linking or let the end >> user do the linking was the same and, hence, the GPL was >> violated. DS> I'll readily concede that my interpretation may be flawed, but DS> this example doesn't seem to contradict anything I said. This DS> binary module was clearly (in the opinion of the FSF) a DS> derivative work of something that was GPL, and hence the DS> company was violating the GPL by distributing the binary DS> module under a license other than the GPL. Well, the question is whether the binary module is a derivative work. Note, the GPL makes four (4) references to derivatives and derivative work. The one most relevant in this discussion is probably the one in paragraph 0.: [...] The "Program", below, refers to any such program or work, and a "work based on the Program" means either the Program or any derivative work under copyright law: that is to say, a work containing the Program or a portion of it, either verbatim or with modifications and/or translated into another language. [...] So they are actually referring to copyright law to define what a derivative work is, and this law might be different in different countries. Thus, whether you violate the GPL or not may well depend on where you are located. Another question is when does the work becomes a derived work? I believe few will disagree that at the moment the two components are dynamically linked, a derivative work is produced. Whether something is a derivative work before the linking, is a debatable question. But, as I wanted to illustrate with this example, even if you make the user do the linking, in that moment a product is created that violates the GPL and the user looses all rights to the GPL part of this new product (at least that is may understanding). Hence, a user can (be made to) violate the GPL. DS> Also, as far as I can tell, your description applies to the DS> situation with Nvidia's binary kernel drivers for the Linux DS> kernel (which is GPL and not LGPL AFAIK), A simple google search would have confirmed to you that the linux kernel is developed under the GPL. There are actually reports that the developers are currently discussing to move to GPL 3 (a bit strange, since GPL 3 is, AFAIK, open to discussion but not yet released) and many of them wanting to stick with GPL 2. (Another thing I find strange, because given the standard clause, one can take GPL 2 code, modify it and then release the new version under GPL 3 [or later].) DS> which is obviously tolerated, so there must have been some DS> other nuances. It is, but there are problems. I have computers with an Nvidia card and run Debian. Quite often, when the kernel image pacakge is updated (even if the same version numbers are kept!), my system breaks and until I recompile the interface to the Nvidia binaries, I cannot use X. Interfacing proprietary binary modules to linux seems to be a perennial problem. Recently (0.5-1 year ago), there were some reports on relevant sites that somebody who was providing support for binary drivers of some digital camera got very upset with how it was made harder and harder for him to provide this support. And when some code was removed from the kernel which made it possible for him to provide the support (and the argument for the removal was that the code only serves this purpose), he really spit the dummy and asked that his code and any other code that he had provided to the linux kernel be removed. As far as I remember, Linus complied but a huge discussion ensued on whether he was actully allowed to withdraw all his code. I did not follow this incidence though to see how it was eventually resolved. DS> In any case, this is the complete opposite of the situation we DS> were originally discussing: there one wants to distribute a DS> GPL-d module that possibly links into a proprietary system. As DS> far as I can tell, the example you quoted above has no DS> relevance in this situation. First, the e-mail that you are answering to was written in reply to your e-mail claiming that a user cannot violate the GPL. IMO, to make this statement true, you have to be careful in how you define user. The e-mail was not written as a reply or input to Spencer's e-mail. Secondly, I have admitted already on this list, and I am happy to admit it again, that English is my second language. But I am quite sure that my command of English is not so bad that I completely misunderstand the point of a discussion. So I have to wonder to what you are referring to as "the situation we were originally discussing". For my part, Spencer was asking (among other things) for opinions on his interpretation of the GPL with regard to the "glmmADMB" package. This package provides binaries that are called (not even linked) from a GPL-d program. And I offered Spencer some of my thoughts on this subject. There is nothing in the "glmmADMB" package (as far as I can tell) which links into a proprietary system, I did not see anything in Spencer's e-mail that was asking about the scenario that you describe and I did not comment on such a scenario. Best wishes, Berwin ______________________________________________ R-help@stat.math.ethz.ch mailing list https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide! http://www.R-project.org/posting-guide.html