(MJ Ray)
and it also contradicts with "No Discrimination Against Fields of Endeavor" because no topic of a secondary section can used as the main purpose.
(Matthew Wilcox)
I don't think that's an interesting case though. Why would you take a document that has nothing to do with a particular subject and turn it into a document that has that subject as its main purpose? That seems ludicrous to me. Put another way: why is that a freedom you want to have?
This is one of the most critically important freedoms of all: the freedom to adapt for a different purpose.
Suppose that Mozilla code was only licensed for use in web browsers. Wham, we lose freedom. Suppose that Apache code was not licensed for use in version control systems. Wham, we lose freedom, and Subversion would be very unhappy. Suppose glibc code had not been licensed for use in compilers or debuggers. Wham, we lose freedom, and lots of valuable cross-pollination with GCC and GDB would have been impossible. Suppose the GCC manual was not licensed for use in essays on the economics of free software. (It actually provides some great examples of funding methods, and quoting some of the sections on various features to go with the information on how they were funded would be quite useful.) Wham, we lose freedom. Oh, wait -- it isn't licensed for such use. Suppose the Emacs manual was not licensed for use in essays on the value of free software. (It would be quite useful, especially in showing examples of how free software processes create better documentation.) Wham, we lose freedom. Oh, wait -- it isn't licensed for such use.
Finally, the hard part - when to use fdocg and when to use DFSG - doesn't seem to be covered at all yet.Indeed.
Unfortunately, it's barely worth even discussing these "dfdocg" until that's dealt with. :-P
For emacs, the "self-documenting text editor and kitchen sink" (emphasis on self-documenting), clearly the manual is part of the program and subject to the DFSG.
Texinfo files are programs intended for interpretation by the TeX interpreter (or the makeinfo compiler and the info interpreter). Presumably they should be subject to the DFSG.
Unless the hard part is done, I would argue that the "dfdocg" would mean next to nothing, because nearly all docs shipped in Debian would still be subject to the DFSG!
We need to start figuring out what our position is on docs. Right now, it's simply "everything is software" which really irritates me (and several other people).
But everything in Debian *is* software. :-) Any sequence of bits is software. This argument has, of course, been at the core of all the recent complaints and disputes. I think most Debian developers understand the traditional meaning of "software" (rather than the degenerate usage which equates it to "computer program"), though I could be wrong. However, we have the opinion of the DFSG author on our side, and the supermajority vote on the editorial changes (which were actually intended as a clarification for people like you who use the degenerate meaning!), and the majority of debian-legal....
And *most* importantly, nobody on the "treat non-programs differently" side has been able to come up with clear, sane tests to differentiate programs (or documentation) from other software.
(In contrast, we *do* have a clear test for what distinguishes license texts from other software, and that's an important reason why most people are willing to have different standards for license texts. If the "Invariant Sections" of the GFDL were included in the license text, rather than having to be part of the main body of the document, with mandatory title entries in the table of contents and so forth, they'd probably squeak under the wire.)
The "Everything is software" view (more accurately "Everything not hardware is software") is *not* an accident; it was arrived at after a great deal of discussion and analysis. Please don't ignore that analysis.
If you do promote this, I strongly suggest that you simultaneously change the DFSG to the "Debian Free Program Guidelines", for clarity. (I, in contrast, would prefer to remove "program" from the few places it appears, and make them definitely "Debian Free Works Guidelines". And add an explicit license text exception.)
Apologies for the stupid mailer; not sending from my usual machine; I had to respond to this when I saw it.