Don Armstrong <[EMAIL PROTECTED]> wrote:
> On Fri, 07 Jul 2006, Joerg Schilling wrote:
> > Andreas Barth <[EMAIL PROTECTED]> wrote:
> > > Also, as usual, you are ignoring the vital fact that the
> > > combination of CDDL and GPL is something between legally dubious
> > > and illegal. Of course, you ca distribute whatever you want, but
> > > Debian is bound to legal behaviour.
> >
> > It seems that you did never read the GPL and the CDDL in depth
> > enough in order to under stand either of them...
> >
> > The GPL explicitely allows to use the code and it only forbids to
> > use GPL code in non-GPL projects. So it is obvious that the GPL
> > allows to use non GPL code in a GPL project.
>
> The GNU GPL only allows this when it is possible to satisfy the
> conditions of the GPL for the distributed work. For example, this is
> why it is possible to combine MIT licensed works with GPLed works.
Your assumption is made on wrong general prerequisites.
The CDDL definitely does not have any requirements on "other" code
as it is a clearly file based license. For this rerason, your statements
(below) on the CDDL are wrong.
What you mention about the GPL is only true in case you put GPLd code
or parts of GPLd code into a non-GPL project. The relevent part of the
GPL is:
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.
Other claims often made about the GPL when talking about the GPL cannot
be found in the original GPL text, viloate the law and thus are void.
The GPL includes no text that is related to GPL projects that include/use
non-GPLd code. As the GPL in general permits to use the code, this is
a permitted use.
Before writing more, it seems to be iomportant to mention a common
missconception:
Both, the CDDL and the GPL are _source_ licenses.
They both _allow_ binary redistribution under certain conditions but it
is definitely wrong to even think about: "under what license might the
resultant binary be".
There is no "binary license for the project" but there is a permission to
distribute/use binaries under certain conditions.
The CDDL enforces contidions under which the resultant binary may be
distributed but it does not enforce _anything_ on the non-CDDL source.
The GPL enforces other contidions under which the resultant binary may be
distributed but it does not enforce _anything_ on the non-GPL source.
Note that you simply cannot create a license that tries to enforce conditions
on other people's code because this would be illegal.
Many people forget about the law and about the original intention of the
FSF when talking about the GPL. Unless you asume that the FSF is acting like
Microsoft, it is obvious that the intention of the FSF it "only" to prevent
the GPLd code from disappearing in CSS projects and to keep it free.
Both do not apply to code under the CDDL because the CDDL is a free license
that itself tries to prevent the code from being made non-free.
> Allow me to make it abundantly clear why the CDDL and GPL are
> incompatible:[1]
>
> CDDL 3.1 requires that the Source Code of Covered Works made available
> in Executable form be distributable only under the CDDL; CDDL 3.4
> disallows additional restrictions. CDDL 6.2 (patent retaliation) is a
> restriction not present in the GPL.
See above: The text is correct, the conclusion is wrong.
The CDDL is a file based license and does not enforce any restrictions on
non-CDDL code.
> GPL 2 requires all of the work when distributed together to apply to
> the GPL. GPL 6 dissallows additional restrictions. GPL 2c is a
> requirement not present in the CDDL.
See above: The GPL (see GPL § 2b) only requires the whole work to be
distributed under the GPL in case that a non-GPL project tries to use GPL
code. The GPL does not contain any text that could cause the assumption the
GPL tries to enforce restrictions on non-GPL code.
> As you can see, they're incompatible with eachother in either
As you see, you just did make the wrong conclusions....
> direction. Indeed, I've been told by those involved in the drafting of
> the CDDL that this was done by design. [See the video of the Solaris
> discussion at Debconf 6 if you want to see someone talk about it; you
> can also see me discussing this issue and others as well in the same
> video.]
This is of course wrong - sorry. Please stop distributing wrong claims.
I have been involved with the creation of the CDDL and I know that your claim
is wrong.
The reason for not using the GPL for OpenSolaris is simple: The GPL (if used
for OpenSolaris) would not allow Sun to create the "Sun Solaris Distribution"
from the OpenSolaris sources.
I had a 1.5 hour phone conference with the lawyer who created the CDDL and the
Solaris chief engineer and we did discuss all aspects of the CDDL and most of
the changes found in the second CDDL contribution to opensource.org (made in
January 2005) are a result of my demands on the CDDL that it needs to be a
world-wide usable and reusable (for private people) license.
The reason why some Sun people believe that the GPL is completely incompatible
with the CDDL is a result from reading wrong FAQs.
I should mention that not all CDDL/GPL combinations are possible and that a
European Author has the right to create more legal combinations than a US
Author has. This is a result of the archaic US Copyright law. This does
however not limit the re-distributability of the code as the USA accept the
European "Urheberrecht" (which is much more than just the US Copyright law)
if the Author is European and European countries accept the US Copyright law
if the Author is a US Citizen.
> That said, it may be possible for the copyright holder (assuming the
> copyright holder is a single entity) to distribute such a work.[2]
> It's just impossible for anyone else to distribute it without
> separating the incompatible bits.
As long as you are unable to proove this claim, let us asume that it is
wrong.
> As a final note, the "mere aggregation" clause of the GPL does not
> apply in this case, as the works have a non-trivial dependency
> relationship with eachother; they are not merely placed on the same
> media for distribution.
It seems that you did not have a close enough look at the facts of the
"cdrtools" project.
I encourage you to first read the file "COPYING" in the root-dir of cdrtools
and to check which sub-project uses which other sub-project.
"cdda2wav" (CDDL) e.g. uses the sub-project "libparanoia" which is a work from
Christopher Montgomery and me and changed from GPL to LGPL with permission
from Christopher Montgomery.
"mkisofs" (GPL) uses the sub-projects "libschily" and "libscg" (both CDDL)
and "libfsfs", "libfile", "libunls" (all GPL).
All C-based projects are compiled by the sub-project "Schily makefiles"
but this is "mere aggregation" as the "Schily makefiles" do not appear
in the final binaries and as the code could be compiled otherwise.
The license change is the result of 2 months of hard work.
Many people have been asked for their legal impression (even many people
from Debian) and nobody was able to proove with facts that my understanding of
the license issues is not correct.
Let me add a statement regarding re-distribution:
I only own about 40% of the Copyrights on mkisofs and I see no problem with
the combination.
As neither the CDDL nor the GPL enforce any restrictions on other code [1],
somebody who likes to redistribute binaries just needs to follow the rules
of all related source code.
[1] If the GPL would try to enforce restrictions on other code, the GPL would
be a non-free license (see DFSG §9).
Jörg
--
EMail:[EMAIL PROTECTED] (home) Jörg Schilling D-13353 Berlin
[EMAIL PROTECTED] (uni)
[EMAIL PROTECTED] (work) Blog: http://schily.blogspot.com/
URL: http://cdrecord.berlios.de/old/private/ ftp://ftp.berlios.de/pub/schily
