Xavier Chantry
2010-02-07 23:20:22 UTC
I also hope this helps somehow.
---------- Forwarded message ----------
From: Karl Berry via RT <***@fsf.org>
Date: Sun, Feb 7, 2010 at 1:42 AM
Subject: [gnu.org #544172] Fwd: [arch-general] An old, tiresome
discussion: cdrtools vs cdrkit
To: ***@gmail.com
Hello Xavier,
Eben just sent me this summary of his discussion with Jörg Schilling on
the subject of the cdrtools mkisofs. He said that it can be
republished/posted anywhere. When/if you do that, please do so *in its
entirety*. All too easy for things to be taken out of context, so let's
at least get the full version out there.
Hope this helps somehow.
Thanks,
***@gnu.org
Date: Sat, 6 Feb 2010 11:24:23 -0500
From: Eben Moglen
In September 2008, I was asked by Mark Shuttleworth if I could help
Canonical discuss with Jörg the measures necessary to include cdrtools
in Ubuntu. I spoke to Jörg by phone at first to explain my role, and
then by email. I reminded him in doing so that I had agreed with him
about the acceptability of combining GPL'd code with a C-library under
CDDL in order to make the Debian/GNU/OpenSolaris stack called Nexenta.
(I believe this part of our conversation is the source of Jörg's
mistaken statement that I somehow indicated that cdrtools is
non-infringing, although the issue of the "system library exception"
in GPLv2 is completely distinct from the problem presented by the
CDDL-licensed libraries libscg and libschilly combined with GPL'd
mkisofs in cdrtools.)
After speaking to Jörg we began our review of the complete source of
cdrtools, and soon verified that GPL compliance on mkisofs was broken.
We told Jörg that as far as we could see he was the only copyright
holder on the CDDL'd libraries, which he confirmed. In that case, I
pointed out, he could give all the permission necessary to solve the
problem, without any license changes: he simply needed to give
permission as the relevant copyright holder on the CDDL's libraries
for combination with mkisofs and distribution of the binary and source
under the terms of GPL, without any additional restrictions. We
drafted for him the thirty-nine words needed: "You are permitted to
link or otherwise combine this library with the program mkisofs, which
is licensed under the GNU General Public License (GPL). If You do,
you may distribute the combined work under the terms of the GPL."
Jörg disputed our analysis. He argued to us: (1) that the binary
mkisofs is not derivative of the CDDL-licensed libraries because it
merely links to them; and alternatively that (2) CDDL Section 3.6
("Larger Works") permits the combination. He claimed to have German
legal advice confirming him on point (1), but without regard to his
German legal analysis, which differs from that of our German lawyers,
the argument is incorrect as a matter of law, at least in the United
States, and GPL'd mkisofs (whose copyright is at stake) is a US work
to which US copyright law applies under the Berne Convention. As to
his point (2), while CDDL Section 3.6 permits combination with code
under other licenses, it nonetheless requires that "the requirements
of this License are fulfilled for the [combined program]." Since it
is impossible to observe certain requirements of the CDDL while
simultaneously respecting the GPL's prohibition of additional
restrictions (GPLv2 Section 6), the CDDL Section 3.6 permission is
insufficient to allow the combination. Hence the permission we
advised him to grant.
Though Jörg continued to argue that he didn't *need* to grant the
permission, he never explained why, in the face of opposing legal
analysis on behalf of the copyright holders of mkisofs he didn't
*want* to grant a harmless permission that would allow his work to be
included in Canonical's Ubuntu distributions. After weeks of
discussion and many hours of my time and the time of my associate
Aaron Williamson, Mark Shuttleworth decided there was no point in
further fruitless negotiation and I agreed. SFLC was not paid for
its work by any party.
---------- Forwarded message ----------
From: Karl Berry via RT <***@fsf.org>
Date: Sun, Feb 7, 2010 at 1:42 AM
Subject: [gnu.org #544172] Fwd: [arch-general] An old, tiresome
discussion: cdrtools vs cdrkit
To: ***@gmail.com
Hello Xavier,
Eben just sent me this summary of his discussion with Jörg Schilling on
the subject of the cdrtools mkisofs. He said that it can be
republished/posted anywhere. When/if you do that, please do so *in its
entirety*. All too easy for things to be taken out of context, so let's
at least get the full version out there.
Hope this helps somehow.
Thanks,
***@gnu.org
Date: Sat, 6 Feb 2010 11:24:23 -0500
From: Eben Moglen
In September 2008, I was asked by Mark Shuttleworth if I could help
Canonical discuss with Jörg the measures necessary to include cdrtools
in Ubuntu. I spoke to Jörg by phone at first to explain my role, and
then by email. I reminded him in doing so that I had agreed with him
about the acceptability of combining GPL'd code with a C-library under
CDDL in order to make the Debian/GNU/OpenSolaris stack called Nexenta.
(I believe this part of our conversation is the source of Jörg's
mistaken statement that I somehow indicated that cdrtools is
non-infringing, although the issue of the "system library exception"
in GPLv2 is completely distinct from the problem presented by the
CDDL-licensed libraries libscg and libschilly combined with GPL'd
mkisofs in cdrtools.)
After speaking to Jörg we began our review of the complete source of
cdrtools, and soon verified that GPL compliance on mkisofs was broken.
We told Jörg that as far as we could see he was the only copyright
holder on the CDDL'd libraries, which he confirmed. In that case, I
pointed out, he could give all the permission necessary to solve the
problem, without any license changes: he simply needed to give
permission as the relevant copyright holder on the CDDL's libraries
for combination with mkisofs and distribution of the binary and source
under the terms of GPL, without any additional restrictions. We
drafted for him the thirty-nine words needed: "You are permitted to
link or otherwise combine this library with the program mkisofs, which
is licensed under the GNU General Public License (GPL). If You do,
you may distribute the combined work under the terms of the GPL."
Jörg disputed our analysis. He argued to us: (1) that the binary
mkisofs is not derivative of the CDDL-licensed libraries because it
merely links to them; and alternatively that (2) CDDL Section 3.6
("Larger Works") permits the combination. He claimed to have German
legal advice confirming him on point (1), but without regard to his
German legal analysis, which differs from that of our German lawyers,
the argument is incorrect as a matter of law, at least in the United
States, and GPL'd mkisofs (whose copyright is at stake) is a US work
to which US copyright law applies under the Berne Convention. As to
his point (2), while CDDL Section 3.6 permits combination with code
under other licenses, it nonetheless requires that "the requirements
of this License are fulfilled for the [combined program]." Since it
is impossible to observe certain requirements of the CDDL while
simultaneously respecting the GPL's prohibition of additional
restrictions (GPLv2 Section 6), the CDDL Section 3.6 permission is
insufficient to allow the combination. Hence the permission we
advised him to grant.
Though Jörg continued to argue that he didn't *need* to grant the
permission, he never explained why, in the face of opposing legal
analysis on behalf of the copyright holders of mkisofs he didn't
*want* to grant a harmless permission that would allow his work to be
included in Canonical's Ubuntu distributions. After weeks of
discussion and many hours of my time and the time of my associate
Aaron Williamson, Mark Shuttleworth decided there was no point in
further fruitless negotiation and I agreed. SFLC was not paid for
its work by any party.