Richard Braakman <[EMAIL PROTECTED]> writes: > On Fri, Dec 14, 2001 at 09:31:58PM -0800, Thomas Bushnell, BSG wrote: > > I prefer a proportional limit for two reasons. First, a fixed limit > > invites the abuse of splitting a big invariant thing into a bunch of > > packages. Second, a proportional limit guarantees that we get some > > real fully-free documentation along with the invariant text. > > I think it's time for some solid definitions here. The term "invariant", > as used in the GFDL, refers to material that is nonmodifiable and > nonremovable. You would not be able to get around a fixed limit by splitting > up a package, because each partial package would have to contain all of > the invariant text.
Right, but the fixed limit proposal would extend beyond just the GFDL. Perhaps a developer writes a horrid novella, and puts one short bit in each of many packages, marked invariant. They have thus subverted the point of the restriction by splitting their crap into many packages. A proportional test at least requires them to write proportionately that much real worthy stuff, I hope. > I myself have a far bigger objection (in terms of Freedom) to > nonremovable text than to nonmodifiable. I don't mind that the GNU > Manifesto is nonmodifiable, and I think it's appropriate for the > Debian Project to publish that document. But I do not think that > a manual with any invariant text is "real fully-free documentation". > It is documentation that _would have been free_ except for the invariant > text attached to it. Yes, in this discussion I think people have been conflating nonremovable as just one kind of nonmodifiable, but you are right that it is especially worse, I think. > I don't think we should accept any manual with invariant sections > as free; but as a compromise, I would support a policy that identifies > a specific set of texts as acceptable. I'm opposed to any generic > limit based on size. Software projects do merge and share code, and > at some point they may have to share parts of their manuals. This means > that invariant sections will gradually multiply and spread to the manuals > of related programs. We should keep a tight rein on their numbers, and > we should judge based on content as well as (fixed) size. It sounds like you agree with the spirit of Anthony Towns's proposal?
