Re: Debian i386 freeze
Oliver Elphick <olly@lfix.co.uk> wrote:
> Raul's proposition appears to go that the GPL licence, applied to the
> software by its authors, entitles 3rd parties to rights beyond those
> given to us by the authors. I cannot see any possible way to support
> this argument.
No, the client never can get from the seller a right which the seller
doesn't have. However, the client can show that the seller was
grossly negligent.
How might that happen?
One possibility is that the client gets sued for something (negligence,
patent violation, whatever), and the client's want to spread some of
the responsibility around, to minimize the damage from that lawsuit.
> The GPL (applied by kde) can only apply to kde's own code, not to
> anyone else's. If it is necessary to link some non-free software, in
> the knowledge of kde (which is the case), they must necessarily have
> given consent to do this by the very fact of releasing their software
> and inviting the world to use it. In addition, one of their developers
> has specifically given consent on this mailing list.
Yes, they've given permission to link against Qt, and they've given
permission to use the result. But they've not given permission to
distribute the result. These are three different kinds of permission.
> People keep losing sight of the fact that the licence is applied by
> the authors of the software. It gives the right to use the software
> under the conditions stated. The licence is not a contract; it is
> completely one-sided. It cannot be enforced against a supplier by a
> recipient, because it is NOT a contract. It can only be enforced by
> the software authors.
A contract exists where ever something of significant value is exchanged
for something else of significant value (money is a classic example).
It doesn't even have to be in writing (thought it's much harder to
prove an oral contract). So when distribution of software involves
the distributor getting something in exchange, a contract exists,
and the license under which the software is distributed is a part of
that contract.
> Any contract for the supply of a CD containing software is a separate
> issue. Debian do not do this; therefore no-one can have an enforceable
> contract with Debian. Anyone supplying a CD should state that it
> is supplied as a convenient means of obtaining the Debian software
> archive; they should not give any warranty about the software itself
> (after all, how could they), nor should they be claiming to be
> conveying any rights in the software. They should merely offer to
> supply a CD from which the archive can be read without error.
I don't think you've read the GPL, because the GPL says very
explicitly I can't distribute software under these terms.
I can only distribute GPLed software if I guarantee certain rights to
the person who I distribute it to. The GPL says, very clearly,
that if I can't guarantee those rights that that's ok, but that I
can't then distribute that software.
> The purpose of the GPL is to enable software authors to keep their
> software free. It contaminates any derivative works, so that they in
> turn have to be under the GPL. However, it cannot work backwards,
> otherwise Motif would now be GPL'd too!
We should never have distributed GPLed code linked with Motif.
We were wrong for ever having done so, and we should stop as
soon as possible.
I'm sorry.
> This is where the frustration comes in - people do not seem able to
> understand that the licence, whether GPL or otherwise, is wholly under
> the control of the authors. It can be varied or withdrawn at any time.
> This is inherent in the authors' rights under the Berne Convention.
> I think that the convention might well allow an author to withdraw
> software from free use, even if it had previously been put under GPL.
> The author might publish an announcement withdrawing all licences,
> and this would (I think) be legally binding; he would be unilaterally
> withdrawing a right that he had unilaterally granted. Of course,
> he would have trouble enforcing this, because of the difficulty of
> ensuring the announcement was seen by everyone affected.
I don't think there's anything in the Bern convention that says this.
Once somebody else owns something, they own it. If they acquired
it using shady means (or never had a right to it in the first place)
that's one thing. If they have a legal copy, that's very different.
However, note that if you are still the author (and thus have all
rights to the work) you can still release your work under a license
that lets you link it with proprietary code. The GPL doesn't
take that away from you.
But you actually have to provide the other license terms, because
the GPL by itself doesn't grant that kind of permission.
> (If the author had supplied software for a consideration, he would not
> be able to withdraw the licence to his customer, because then there
> would be a contract enforceable by the customer against the author.)
No, if that's the way it worked there could never be public
domain software. [Public domain software is software you give
away for free under terms where someone else can incorporate
it into a product and sell it without giving you any rights
to the result.]
> Any way, the crucial points are:
>
> A software licence, unless part of a contract, is enforceable only by
> the software authors.
>
> Debian do not enter into any contracts.
>
> Therefore, no licence can be enforced against Debian by a licensee.
>
> It follows that, for Debian, there is no problem and no need to change
> anything.
I think it's a bad policy for us to distribute software under terms that
say we're not allowed to distribute it.
For example, you're saying that we are not allowed to have an ftp site
maintainer who has the ftp machine supplied by some local business
in exchange for installing debian on the systems of that business.
Because remember that once the people have access to debian, and apt,
they can fairly easily install any software from the ftp site.
If the license installed put the burden on the person who choose to
install the software, then it wouldn't be such a big deal. But
the GPL has restrictions on how you're allowed to distribute it.
There's nothing unclear here: the license on KDE clearly says
that we can't distribute it without explicit permission. Even
now no one is granting explicit permission. Therefore we don't
have permission.
Otherwise, all of a sudden we go from an independent group of volunteers
doing their best to put together a free system that everyone can use to
a not so independent group of volunteers who are constrained by
the Qt license.
> A CD manufacturer should explicitly disclaim any rights in the
> software on the CD; he should only claim to provide a readable CD that
> contains a specified set of files. But, whatever he chooses to do, it
> is not our problem but his. If he considers it worthwhile, he can add
> a disclaimer at the top level of the contrib CD.
He can't do this with GPLed software. He can't distribute GPLed software
that's incorporated into a work which doesn't meet the GPL's standards
for freedom.
> Now, some detailed analysis of the GPL, as applied to kde:
>
> 1: "You may copy and distribute verbatim copies of the Program's
> source code as you receive it,..."
> ^^^^^^^^^^^^^^^^^
> This indicates that you do not have to make efforts to go beyond what you
> received yourself, when distributing GPL software.
No. This is permission to do something, but that doesn't relieve
you from following the rest of the license.
> 2. b) "You must cause any work that you distribute or publish, that in
> whole or in part contains or is derived from the Program or any
> part thereof, to be licensed as a whole at no charge to all third
> parties under the terms of this License."
>
> This is the basis for claiming that linked libraries `contaminate' a
> software product. This is obvious in the case of static linking, but
> extremely dubious in the case of dynamic linking, which does not
> include any of the library but only a set of hooks for calling its
> facilities. A program that dynamically links a GPL library is not
> derived from it (that term would imply a new library developed from the
> GPL code), neither does it contain it; therefore, as a matter of objective
> analysis, it seems to me that the GPL does not achieve the object of the
> FSF in the case of dynamic linking. [No doubt this is because the GPL
> was drawn up before dynamic linking was in common use(?)]
This doesn't say anything about linking at all.
Furthermore, since the author of the license was a lisp hacker (lisp
offers far more flexibility than dynamic linking does, in terms of
creating distributed works) I don't think that it was written in
ignorance of the possibilities.
Or are you trying to argue that the Qt code is not a necessary part
of kde?
> 3. "You may copy and distribute the Program...provided that you also...
I'm going to just skip over the irrelevant clauses. Yes, you have to
satisfy those clauses to satisfy the GPL, but we'll just assume that
they're easy to satisfy.
> 6. "Each time you redistribute the Program (or any work based on the
> Program), the recipient automatically receives a license from the
> original licensor to copy, distribute or modify the Program subject to
> these terms and conditions..."
>
> `The Program', in this case, means kde, not Qt, which the kde authors have
> no rights over (nor do they claim any). There is nothing here to support
> the idea that this asserts any rights over Qt.
The program doesn't work, in that case, at all. This is a rather
serious bug.
But since many thousands of people are using KDE, I don't think a
court would accept this fiction.
And it is this lack of rights over Qt that requires an exception to
the GPL to allow us to distribute KDE in its current form.
> 7. "If ... conditions are imposed on you ... that contradict the conditions
> of this License, they do not excuse you from the conditions of this
> License..."
>
> This does not apply to kde/Qt, because the GPL is talking about a new
> program made from kde. It envisages some circumstance where a licensee
> wants to incorporated patented algorithms in a derived work, or withhold
> the source to some part of a derived work or where rights given
> by the kde authors are not permitted to be passed on by a licensee.
> The first two do not apply, and all the rights given _can_ be passed on;
> no right over Qt was given, so none need to be passed on.
Well, I'm not sure what to think here. You've told me that KDE
doesn't work at all, and now you're telling me that combining this
non-working code with this other code called Qt isn't creating a
new work.
Why am I reminded of an old game involving three walnut shells and
a pea?
> 10. "If you wish to incorporate parts of the Program into other free
> programs whose distribution conditions are different, write to the author
> to ask for permission..."
>
> Again, this is talking about incorporating _kde_ into other programs;
> it does not apply to kde's own reliance on Qt.
Please, author, give us permission to distribute this code legally?
> It is also clear from the above that kde cannot incorporate or derive
> from any other GPL software without the consent of the authors of that
> software. The kde maintainer should make sure that there are no such
> unauthorised works included in the kde packages.
This, at least, is very true.
--
Raul
--
To UNSUBSCRIBE, email to debian-devel-request@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmaster@lists.debian.org
Reply to: