> In the case of GNU Emacs, for example, there is a problem. Code has > been copied from the program (GPL'd) to the manual (GFDL'd). The > manual is a derived work of a GPL'd work, but is distributed only > under the GFDL. The FSF may legally do this *only* because they own > the copyright on all of the code in Emacs and on all of the text in > the manual. Nobody else has the freedom to modify the system of > code+manual in the ways that the FSF routinely uses.
Actually, that's not entirely true. To the extent that a chunk of published code is purely functional and lacking in "creative expression", or meets either the "de minimis" or the "fair use" standard of affirmative defense, it can be copied into a document distributed on any terms you like. IANAL, etc., etc., but there is plenty of precedent on the limits of the reach of copyright law with regard to the functional aspect of source code. For a view from a different angle -- the status of source code with respect to First Amendment protections -- see Berstein v. USDOJ (withdrawn) and Junger v. Daley (good law), discussed and linked at http://www.discourse.net/archives/2003/10/_the_bernstein_cryptography_case_is_dismissed.html . Both courts ruled that source code has both functional and expressive aspects, and it's not much of a reach from there to find that it is the expressive aspects that can be protected by copyright. There's no need to make that leap, though; the classic case dividing the functional from the expressive in a copyright setting is Sega v. Accolade 1992 ( http://www.eff.org/legal/cases/sega_v_accolade_977f2d1510_decision.html ). The amendment to the decision is particularly interesting, in that it applies both the functionality and de minimis criteria to decide that a crucial, arbitrary, small piece of code is not entitled to copyright protection. Cheers, - Michael