Documentation vs. Software (Re: Savannah rejects a project because it uses GPL)

Alfred M. Szmidt ams at
Wed Feb 22 22:00:30 UTC 2006

   > I think you are trying to put documentation and programs into one
   > box.

   No, I'm telling that you cannot put programs and documentation
   shipped electronical in different boxes.

If they are different, then they should be in different boxes.  Maybe
you are speaking about physical boxes?

   > One should always examine what kind of work one is handling
   > before licensing it.  Sometimes it makes sense to use the
   > modifed-BSD license, sometimes the Lesser GPL, sometimes the GPL.
   > Same applies for `text', some `text' is better to be under the
   > GPL, some is better to be under the GFDL.

   The problem with your analogy is, that BSD, LGPL and GPL are all
   free, they only differ in how they protect against non-free stuff.
   The GFDL is different, in that if applied to programs it is simply
   non-free. It can be better for you to license something non-free,
   but I do not want to use it or recommend its usage.

Documentation and software are clearly different, different rules
apply.  Licensing a document under the GFDL does not make it non-free,
it makes it `non-free software', but then, it isn't software at all so
calling it `free software' is equally (in)correct.

   > >  Well, the need for freedoms for program source code is not
   > >  clear either.  Many people distpute that.
   > How are the freedoms for program source code not clear?  I should
   > be able to control what my machine does, what it does, and tell
   > people what it did.

   And change what it does and distribute it to others. But there are
   people telling me that is not needed for programs, and there are
   people like you telling me it is not needed for documention.

Can you explain to me why you require the right to modify what I have
written at all times?  Compared to why you need the right to modify a
program I have written at all times?

   > >  The free software licenses do. The FDL does not allow
   > >  modification in reasonable ways. (plus has some glitches with
   > >  copying).
   > How doesn't it allow for it in a reasonable way?  Last time I
   > checked, I could modify a document in anyway I choose.

   Now, let's take some example. Let's take the documentation of some
   protocol, assume it is only available as FDL. I want to write a
   protocoll sniffer for that protocol, so I take the tables for the
   different events and their structure from the documentation, use
   some editor macros and some hand-editing to make it a program
   source for a package printer. Thus I get source code for a program
   which I have only available as FDL. (And FDL for source is
   unsuitable, as for example I'll have problems getting it into some
   embedded chip with the whole license and perhaps some additional
   licenses in it).

So don't license the data structures, examples, etc, under the GFDL.
Pick the proper license for the different works you use.  No single
license will work for everything.

   A documentation unrelated to a program is useless. If it is related
   it should not be separated.

But they _are_ seperate, documents aren't programs.  You can still
distribute them in the same physical box if you wish, nobody stops you
from doing this.

   It's reasonable to expect stuff will
   move the one way or the other. A good small definition or
   description from the comments of a program is often a nice sentence
   in the documentation.  A good short description from the
   documentation can sometimes be added as commentary in the
   program. Documentation source might include machine readable stuff
   to be transformed by some scripts or even human to code
   parts. Programs might get tooltips built in containing stuff from
   the documentation. Some part of documentation could be rewritten
   using an auto-generated stuff or docstrings or something like this
   from the code. There is exchange between them, and incompatible
   licenses make this impossible. Thus a license for documentation
   needs to be includeable in some free software license to be free.

Yes, and the GFDL allows for this.  Nobody stops you from licensing
your examples under the GPL and the rest of the document under the
GPL, likewise with the code to be licensed under the GPL.  The GNU C
Library manual does this for example.

   > How can more freedom to change what I wrote be good?  Would you
   > be ok if you for example wrote "I Bernhard R. Link like dragons",
   > and then someone comes and changes this (the text is licensed
   > under the GPL or similar) to "I Bernhard R. Link hate freedom"?
   > I don't think anyone has such a right.

   There are laws against slander and other things like that coping with
   your example. But also for manifestos and personal opinions freedom
   is a useful thing. Other people might want to write manifestos too,
   and might want to express their opinion. Having to formulate them
   all from scratch is quite some work, and allowing people to save time
   is in general good. (Of course sometimes you like not to be good
   to persons you do not like, or that want to harm you. I do not
   want to say that such things should be free, only that it is nicer
   if they are.)

There is no law in the world that will protect you from slander if you
explicitly allow for the right to modify a work.  You cannot sue a
person for using a free software program in a manner that you consider
bad since you explicitly gave this right.  Same applies if you give
the explict right to modify a text in anyway or form.

   > Anything that is of a digital nature is sensible to include in an
   > OS.  This does not mean that you, or anyone, should have the
   > right to change for example the GNU manifest to state something
   > entierly different, while still making it sound as if it is the
   > GNU project who wrote it.

   Again, writing something making it soundas if someone else wrote it
   is libel or slander or any other such term layers might use and
   totaly unrelated to copyright.

Plain wrong if the license explicitly allows for such modifications.

   > I find it quite easy: Computer runs program, human reads text.  I
   > would like to see a single human who can actually run a program
   > of sufficent size.

   But humans read computer programs. (And with free software that is
   an important part of programs). And documentation is first read by
   computers, too.

It isn't read by a computer.  It is read by a parser which is read by
a computer.  By your logic, since this text is `read' by emacs, which
inturn is a program, you should have the explicit right to change what
I wrote to something I simply disagree strongly with and claim that I
actually wrote it.


More information about the Discussion mailing list