Path: utzoo!utgpu!jarvis.csri.toronto.edu!mailrus!uflorida!rex!kozma From: kozma@rex.cs.tulane.edu (John Kozma) Newsgroups: comp.misc Subject: Re: The GNU Public License Summary: Some of you people don't understand the copyright law Message-ID: <991@rex.cs.tulane.edu> Date: 2 Aug 89 22:08:27 GMT References: <26@ark1.nswc.navy.mil> <26719@agate.BERKELEY.EDU> <5351@ficc.uu.net> <9655@phoenix.Princeton.EDU> <1989Jul30.210646.12194@twwells.com> <1811@hudson.acc.virginia.edu> <1989Aug2.044845.4211@twwells.com> Reply-To: kozma@rex.UUCP (John Kozma) Organization: Computer Science Dept., Tulane Univ., New Orleans, LA Lines: 72 Distribution: I posted a rather lengthly article in comp.lang.misc, but nobody seems to have noticed it. Hence this follow-up. I think there are a lot of misconceptions about the effect and enforceability of the GNU license, viz: 1) Something is not a "derivative work", as that term is used in the copyright law, just because it incorporates any part of another work. As to the parts copied to the subsequent work, you need to determine whether that constitute fair use of the previous work, and whether the parts copied in fact comprise copyrightable subject matter at all. (Everyone seems to be pretty clear on the fact that the object code produced by a compiler is not a "derivative" or a "compilation" of the compiler, as those terms are used in the copyright law.) 2) A person who offers you a contract, such as a software license, cannot predicate your acceptance of that contract on anything he chooses. In particular, I don't think you can be said to accept the terms of the GNU license just because you "use" software covered by it. This is because a copyright owner does not have an exclusive right of use in the first place. Thus, anyone who obtains a copy of a program, whether or not it's a legal copy, has the right to use it, regardless of what the copyright owner says. (Of course, if you accept a contract by signing it, that's a different situation.) 3) A public license appended to a copyright notice may, as is frequently done, include restrictions on copying and subsequent distribution of copies, such as including the same notice on any such copies which are made and distributed. This does not automatically imply that the same restrictions may be made on copying and distributing derivative works. The public license, presumably, may give the right to make derivative works under certain conditions which are necessarily fulfilled before the act of making the derivative work is undertaken, or which involve copying and distribution of the parts of the underlying copyrighted work which are incorporated in the derivative work. But keep in mind that the derivative work, if made with the permission of the copyright owner of any underlying work, is itself copyrightable, and to acquire any exclusive rights in that copyright, you must get a signed document from the creator of the derivitive work. The bottom line is that, to the extent that the GNU license purports to allow the creation of derivative works, and then purports to restrict the rights of the owner of the copyright of those works, its validity is highly questionable. 4) If a copyrighted work couldn't be used without copying portions of it, the general result (before the days of software copyrights) was that the portions in question were not copyrightable subject matter. This doctrine, if applied blindly, would prohibit copyrighting of software altogether, since a program can't be used without copying it into RAM. Rather than reach this drastic result, I think the general view now is that there is implied permission to copy a work to the extent necessary to use it. This does not imply that the former doctrine is dead, however. For example, if you use a compiler to produce assembly language, arguably that language is copied from somewhere in the compiler. But the copyright on the compiler does not extend to the tables of mneumonics for output code, because you can't use the compiler with out copying them. By the same token, I think a very good argument could be advanced that various templates, tables, etc. buried in the code for higher level software development tools are likewise non- copyrightable subject matter. On the other hand, I wouldn't rule out "look and feel" copyrights on this basis. 5) I don't think there is any way the GNU copyright owner could object to the practice of using GNU CC to create object files, which are not derivative works of the compiler, and distributing them without source code to ultimate users who link them to GNU library routines. The opinions above are mine-- I am solely responsible for them. Of course, anyone and everyone who wishes is free to adopt them. John P. Kozma kozma@comus.cs.tulane.edu