Re: The invariant sections are not forbidden by DFSG

[Frank Küster <frank@debian.org>]

Anton Zinoviev <anton@lml.bas.bg> wrote:

> On Tue, Jan 31, 2006 at 08:59:08PM +0100, Frank Küster wrote:
>> 
>> >> That makes more than 20 pages of invariant sections, or less than 13% of
>> >> interesting material.  Do you agree that the GNU Emacs Manual is non-free?
>> >
>> > It is free.  20 pages do not obstruct the users to exercise their
>> > freedoms.
>> 
>> I believe that 12 times 20 pages do obstruct the freedom.
>
> If you have 12 documents each with 20 pages invariant sections and all
> invariant sections are different then this would indeed result in 12
> times 20 pages in the combined document.  For a printed document that
> would be impractical.  (I'm sorry that I didn't come to that
> conclusion from your previous posts but this situation appears to me
> very unlikely.  In most real cases GFDL won't cause that many
> invariant sections.)

Why do you think it is unlikely?  It has also been discussed, and
there's even a GR amendment, whether the GFDL without invariant sections
should be considered free, and here we are talking about problems with
invariant sections.  So let's assume that we are talking about documents
that in fact do have invariant sections.  I don't know of any but the
FSF ones, and I guess the length of the invariant sections should be
pretty similar in each.  So I have no data about the average length of
invariant sections in non-FSF GFDL'ed documents, but why should they be
much shorter?

If I combine only FSF-copyrighted material I can of course merge most of
the invariant sections, but not if there are different invariant
sections from different copyright holders.  But a license that's only
free if I use it together with works from the same copyright owner isn't
free at all.

> It is always a great inconvenience to be unable to combine two or more
> free works into one.  Nevertheless this can not be a reason to
> consider these works non-free.  For example, if the licenses are
> incompatible then it is impossible and illegal to combine such works.

That's correct, however all these licenses are compatible with
themselves:  There's no problem combining different works with different
copyright that are under the same license.  This is different with the
GFDL. 

> Notice however that we accept as free some licenses that do not allow
> combined works in principle -- that is you can have two works covered
> by same license and yet, you are not allowed to combine them.  An
> example of such license is the Q Public License (QPL).

As far as I know, whether the QPL is free is under debate, but I usually
don't follow -legal.

> The sources of
> all derived works should be distributed in the form
> original_source+patch so if you have two works covered by QPL then
> there is no permissible way to distribute the source of the combined
> work (unless the combined work is merely aggregation of independent
> derivatives of both works).

Whether this is possible is a technical question, and it has been
explained a couple of times in this thread that patch clauses are
different from invariant sections.

>> >  (Although it can be forbiddable if you want to donate large
>> > quantity of printed documents to your students.)
>> 
>> So the freedom to give away documents to students is not important, or
>> what? 
>
> The convenience to give away copies is important but not something we
> can use to determine whether some work is free or not.  In some cases
> you can give only 10 copies without much inconvenience, in other cases
> you can give 100 copies and in other cases you can give more than 1000
> copies.  The only difference is how big the number is and the license
> is only one of many factors that have influence on that number.

I don't think this is correct.  If the signal-to-noise ration is only
13%, then it's only 13%, no matter how often I copy or print the
document.  

Regards, Frank
-- 
Frank Küster
Single Molecule Spectroscopy, Protein Folding @ Inst. f. Biochemie, Univ. Zürich
Debian Developer (teTeX)