The idea of numbering from 0 (in nonprogramming contexts) as a sort of
hacker shibboleth appears to be a 1980s joke that is rooted in an
erroneous overgeneralization from C syntax. I am prepared to revert
this change if someone can demonstrate for me that "counting from
zero" has some sort of techno-cultural basis outside of traditions
established by C programmers.
At least some proto-versions of the GNU GPL (preceding GPLv1) began
section numbering from 1. See, e.g.:
http://www.free-soft.org/gpl_history/emacs_gpl.html
To my knowledge, all FLOSS licenses outside the GNU license tradition
with numbered sections number them from 1, not 0.
I have eliminated the definition of "Development Activities", which
was used a bit carelessly. The result is a bit wordier but otherwise
an improvement.
I firmly believe there should be an AGPL-ish variant of
copyleft-next. However, using language from AGPLv3 as placeholder
seems of no value. I am convinced that drafting the supplementary
license language that will be the AGPL-ish variant of copyleft-next
requires starting from scratch.
It's certainly historical GPL policy that Separate Works (if we can get an
adequate definition of Separate Works) need not be included as part of
Corresponding Source.
However, it's industry practice and is believed by most to be required
under GPLv2 to at least name, identify, and describe Separate Works that
are needed.
For example, for decades, the expected requirement under GPLv2 for
Microsoft Windows software has been to describe the precise proprietary
compiler suite (and its specific version number) needed to build the
software.
This change implements that long-standing policy and interpretation into
clear license text for copyleft-next.
Many of my various proposed improvements to Corresponding Source
definition were rejected, and the Corresponding Source definition has
drifted considerable since then.
This change reincorporate important verbs and nouns into the definition
that are necessary to avoid arguments by violators that they can provide
inadequate build/install scripts and/or instructions.
Having "skilled developer" in there is just a pure give away to
violators. They'll argue that our developers aren't skilled enough, so
inadequate CCS is acceptable. There is no way that we should just give
that away to them.
Again, violators could easily say that what they have are not instructions
per se, even if we as copyleft enforcers would disagree. Adding
"information" here makes it much more general than just scripts and
instructions.
For good measure, an "and/or" is put in place as well.
I don't like "specific" here. I see lots of CCS that has "vague
instructions" which are nonetheless helpful. We don't want violators to
be able to say: "Well, the instructions we have aren't very specific, so
we don't have to provide them." If all they have are vague instructions,
they must nonetheless provide what they know and what they've got.
Violators *will* argue this if we don't remove "specific". So, it's
herein removed.
There is no reason in my view for Corresponding Source to only have
meaning when a Covered Work is Distributed. Therefore, remove the word
"Distributed" from the Corresponding Source definition.
I don't see a specific difference between "all parts of the Covered Work"
and "all of the Covered Work". Since Covered Work is a defined term, if
anything, "parts" should be included in its definition. However, I tend
to believe it just isn't necessary. "all of the Covered Work" means "all
parts". How could it not?
'Source Code' and 'Object Code' definitions are moved up to section 0.
This then enables simplification of section 8. It now begins with the
definition of Corresponding Source; the definitions of 'Product' and
'Minimum Period' are rolled into subpart a.
Justification: it is already obvious that negligence is a form of
tort. If there is some legal value in making the inclusion of
negligence explicit, we can revisit this change.
The sentence "You may have additional rights under applicable law
which this License cannot change." is deleted. This was adapted from
no-warranty provisions in certain Microsoft licenses. The deletion is
justified by the fact that it merely states and at most provides
notice of a legal truism. If there is some significant legal benefit
to doing so, we can restore it.
This change is influenced by my recent LinuxCon talk in which I argued
that the OSI's Open Source Definition is rather unsuitable as a
normative definition of open source given its peculiar historical
roots in the mid-1990s concerns of Debian for its downstream
commercial redistributors.
I do not understand why it ought to be necessary to speak of "shared
libraries and dynamically linked subprograms" and assume it is
sufficient to speak of "shared libraries".
The main part of this change is to replace the expansive set of verbs
used in the initial part of the definition of 'Corresponding Source'
(similar to the corresponding definition in GPLv3) with something more
like the definition used in GPLv2 section 3.
'Relicense' may not be sufficiently precise to interpreters of the
license who are well outside of FLOSS legal culture. The defined term
'Distribute' is a good substitute.
Deleted 'for commercial gain' limitation, in response to criticism
from Luis Villa and Mike Linksvayer. Added OSI OSD as basis for
exclusion from proprietary license category.
The issue here is, in a sense, one of proof from the licensee's
perspective. How will the licensee know whether the proprietary
commercial version really would be a 'Covered Work had you prepared
it'? Hence my introduction of language of "reasonable belief" in the
previous commit. However, I now think it is preferable to use more
absolute language, to ensure adequate deterrence without
unintended overdeterrence.
Prior to the recent commit replacing the provision with a [FIXME], the
anti-anti-circumvention provision read:
Each Distributor of a Covered Work (i) waives all rights to forbid
circumvention of technological measures achieved by exercising
rights under this License with respect to a Covered Work, and (ii)
disclaims any intention to limit operation or modification of the
Received Work as a means of enforcing such Upstream Distributor's or
third parties' legal rights to forbid circumvention of technological
measures.
which is descended from the second paragraph of GPLv3 section 3.
I am now deleting this provision in its entirety because I do not see
what real-world scenario it remedies. I assume that copyleft-next is
likely to be used principally as a license for software, much like the
GPL family.
1) Waiver: Read 'Covered Work' as broadly as possible (under
copyleft-next, at least, this means either the particular work
received by the licensee or any 'Derived Work' of that Received
Work). (Note that currently copyleft-next does not define 'Derived
Work' such that it cannot be a lateral 'Derived Work' prepared by some
other licensee in some (possibly different) distribution chain
beginning with 'Us'.)
For this waiver to be meaningful, we would have to suppose that a
Distributor (or 'We') might decide to invoke anticircumvention civil
remedies against some third party exercising copyleft-next-granted
copyright licenses "with respect to" some Covered Work. This could
mean either that the underlying copyright-protected work is the (or a)
Covered Work, or perhaps some unrelated work copyrighted by the
Distributor/We in which the asserted effective TPM is a Covered
Work.
Are these scenarios *possible*? Yes. Do they seem so likely to justify
addition of a section of the license to deal with them? At the moment
my response is 'no', but perhaps there are realistic examples I am
overlooking.
2) Disclaimer of Intent: My justification of the deletion of the
language descended from GPLv3 section 3 paragraph 1 applies here as
well. It is not clear that a 'disclaimer of intent' does anything at
all. I am not aware of anti-circumvention law statutes that give
weight to the intent of a distributor of technology that is used in a
TPM.
We will restore this provision (in improved form) if people come
forward with information suggesting that the waiver and/or disclaimer
of intent are worth restoring.
But if the only purpose of these provisions is, say, a political
statement of opposition to anti-circumvention laws, or, say, the fact
that they form an important addition to GPLv3 not present in GPLv2,
that is not a good enough reason to keep them in.
I am aware that some of my logic above might be applied by some to
other provisions, such as the patent license grant. But the difference
there is that there is now more or less a consensus that FLOSS
licenses should have some sort of patent license grant, however
minimal or maximal. No FLOSS license other than the GPLv3 family has
attempted explicitly to address anticircumvention laws since they were
introduced.
Finally, it seems to me that the 'No Further Restrictions' clause and
the copyright license grant ought to provide some of the safeguards
that the anti-anti-circumvention provisions have aimed to achieve;
and, as with some other deletions of GPLv3-specific provisions, the
outbound compatibility with (A)GPLv3 means that copyleft-next code
should satisfy those who lament the deletion of these provisions from
copyleft-next.
