[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: PROPOSED: interpretive guidelines regarding DFSG 3, modifiability, and invariant text

On Sun, Nov 25, 2001 at 09:11:36PM -0800, Thomas Bushnell, BSG wrote:
> I think I agree with your goal here completely.  For that reason, we
> should not make any kind of statement that "everything licensed under
> the GFDL is ok".

Agreed.  I will continue to assert that there is an extra level of
scrutiny required for GNU FDL-licensed material that is inapplicable to
GNU GPL-licensed material.  If all copyright within a GPL'ed work is
consistent and accurate, then it's fine; it is impossible for an
author/copyright holder to misapply the GNU GPL to his own work.  The
same is not true of the GNU FDL.

> *Somebody* will always need to look at the exact
> details and then make a judgment call.  I see your proposal as a way
> to try and automate the judgment call; I would rather we had a policy
> (if we need one) that gave guidance on that judgment call.  

No, I'm not trying to automate the judgement call.  People still have to
read/view the material and determine the proportion/amount of Invariant

My policy is in fact intended as guidance.  It is rather specific
guidance, but guidance nonetheless.

> Oh, I could certainly accept many!  But it just takes time to bear
> with me, and I do listen and learn.  I think, for example, that I'm
> coming around to seeing some kind of "official guidance" as a
> reasonable thing, so that debian-legal can offer a fairly clear
> guideline for developers to follow.

That's all I'm trying to accomplish.  Again, I reiterate that my
proposal has no particular binding force beyond whatever consensus can
be reached among the developers and (particularly) the archive
administrators, as they are empowered to reject packages from inclusion
in the Debian package archive.

> > You are asking us to allow a lot more non-modifiable, and thus non-Free,
> > material into Debian than we currently do.  I find this stance
> > unacceptable.
> Really?  I'm stunned that you could say this.

I don't see why.  It is pretty obvious to me that the existing DFSG
provides no exceptions to clause 3.  The work must be modifiable and
modified versions must be redistributable under the same license as the
original.  Period.  It doesn't say "except for the license text itself".
That is a de facto exception that Debian has made in the past.  As far
as I know, we have made no others, except by accident.  It is also true
that non-free material sometimes "sneaks" into main (usually through
ignorance of its existence), and changes have to be made to packages to
exclude it.

> Any packages consisting of mostly non-modifiable text would not have
> such portions be "small relative to the size of the package".  That's
> a very important part of my paragraph.

The sum total of such non-modifiable material could end up being
considerable, however, and would certainly be higher than the amount
that is presently permitted de jure.

Again, I'm baffled at your astonishment.

> > What kind of crisis do you require to justify action in your
> > opinion?
> I think if we found that there were a bunch of developers confused
> about the issue, then it would be good to work out something
> concrete.  So far, that just hasn't happened.  

I think it's pretty obvious that there is at least some confusion on
this issue.  I've seen at least three confused people post on this
subject to debian-legal already.

> So, you see, we are closing in on issues.  :)

Fortunately, you have eroded my grounds for pessimism a little bit.  :)

> 1) All software must permit modification.
> 2) All text which is documentation of the software must permit
>    modification.
> 3) Licenses and similar things quite reasonably do not permit
>    modification.

I'd strike "quite reasonably" from the above.  It is not obvious to me
that there is some moral imperative for the FSF to not permit
modification of the terms of the texts of their licenses regarded as
standalone entities (this is not the same thing as altering the text of
a license *as applied to a specific work*).  I do believe I understand
why the FSF does so (as a strategic measure against license
proliferation and the consequent compatibility issues), and I'm willing
to accept this strategy as a means to a greater end, but that doesn't
mean I think the practice is justified in its own right.  In fact, in
the general case I find the practice abhorrent.  I have a music CD with
a notice on the rear liner that says "the text of this liner is subject
to copyright protection", or words to that effect, and all that's
printed is the name of the artist, the album, the song titles (which,
incidentally, are not copyrightable in the U.S.), some brief credits,
and other copyright and licensing notices!  I find this particular case
highly offensive and morally bankrupt, and I perceive only a difference
in degree, not in kind, between this and copyrighting the text of a
license or contract.  Copyright exists solely to promote the progress of
the useful arts and sciences, and is only legitimate insofar as it does
so.  I do not interpret that to mean that copyright should be
reflexively (in the logical sense) applied.

Nevertheless, I won't belabor this point further, as I'm personally
willing to grant the FSF, and other free software licensors, latitude on
an ends-justifying-means basis.

> 4) We would prefer it if other text not falling in categories (1),
>    (2), and (3) also permitted modification, but we recognize that
>    many of them historically have not (the Manifesto chapter of the
>    Emacs manual, for example).  We think that it's OK for Debian to
>    include such things, but we are nervous and sketchy about it.  We
>    think at the least the following should be true:
>    a) No bending on principles (1) and (2).
>    b) Any such unmodifiable text must be a small portion of the
>       package.

I agree with this, however I think you need to consider the possibility
of things other than text being held unmodifiable.  Shall we permit that
or not?

> And I'm willing to go along with the creation of some kind of clear
> debian-legal advice to developers encapsulating these principles.

That's my goal.

> Where I think we do still disagree is about:
> A) Whether we should codify a specific numeric standard for "small
>    portion",

I believe we should.  Recall my fondness for bright-line tests.  :)

> B) What that numeric standard should be, if we want to codify it;

I'm quite open to suggestions on this point.  I pulled 16kB or 5% out of
the air.

> C) Whether the numeric standard should vary depending on exactly what
>    kind of text is involved.

I remain flexible on this point as well.

Furthermore, I support in principle the granting of waivers to this
policy on a case-by-case basis.  However, such waivers should require a
very high degree of consensus, and there should perhaps be some absolute
threshold of non-modifiable content beyond which a package cannot go.
Moreover, I'll be quick to withdraw support if I detect abuse.

> Is that a fair summary of "the show so far"?

I believe so.

G. Branden Robinson                |
Debian GNU/Linux                   |       kernel panic -- causal failure
branden@debian.org                 |       universe will now reboot
http://people.debian.org/~branden/ |

Attachment: pgpHjPFUjWwiG.pgp
Description: PGP signature

Reply to: