On Fri, 2002-08-30 at 18:44, Nick Phillips wrote: > On Tue, Aug 27, 2002 at 01:19:35AM -0500, Branden Robinson wrote: > > > > While the copyright holder can certainly distribute obfuscated source > > and no one can tell him not to, the GNU GPL by which the licensees > > (i.e., we, and our users) are bound requires them to distribute source > > in the "preferred form for modification", which is not available from > > the copyright holder. Licensees cannot meaningfully fulfill the > > requirements of the GNU GPL, therefore they "may not distribute the > > Software at all". > > I think you're a little over-zealous in your interpretation. The original > distributor is clearly the only entity not distributing what for them is > the preferred form for modification, and that's their prerogative. Whilst > we may not like it, I don't think it in any way makes it undistributable; > as far as we are concerned their commented sources don't exist -- to us, > the obfuscated source *is* the preferred form for modification.
Consider the case where a GPLed program is distributed with .o files that are linked in at link time. The author could say, under the same logic and with a straight face, that the .o is "the preferred form for modification". C source code is not "special" or "magic"; it is not metaphysically classified by God as source code. The question as to whether the code is source is properly answered by referring to the definition in the GPL: does anyone really modify the particular file directly when they want to change the program's behavior? In the case of obfuscated source, the answer clearly is "no". Thus, the obfuscated C file is "object code", no matter what language specification it technically adheres to. Now, in this case, there's no real foul; the author has just chosen a contradictory license, which means that no one really knows what the distribution conditions are. If there were third-party GPL code included in that program, however, the third party could demand the "preferred form for modifying" those C files; if the case made it to a courtroom, the onus would be on the author to prove that (s)he really did modify those obfuscated C files when (s)he wanted to change the behavior of that code. > I believe the relevant parts of the GPL are intended to prevent degradation > of the usefulness of sources as they are passed 'down the chain', not at > any point before it even enters the chain (as the GPL has no clout there). So you think RMS intended to allow obfuscated source with the GPL? That's an awfully difficult claim to prove.