alex at alex.org.uk
Tue Sep 6 08:39:33 UTC 2005
--On 05 September 2005 21:52 -0400 "Rod Dixon, J.D., LL.M."
<roddixon at cyberspaces.org> wrote:
> Certainly, you cannot
> share code under the GPL with code licensed under an incompatible
> license, but that is the clear intent of the GPL. Hence, if the GPL is
> viewed as burdening code sharing, then we are no longer referring to
> license proliferation; instead, the project, as someone else has noted,
> is to *tier away* licenses that are copy-left.
I agree with what you are saying from a logical point of view, but I thus
come to a rather different conclusion.
IE, if the aim of the exercise is solely to reduce burden on code sharing,
you end up "tiering away" all licenses that are copy-left (reciprocal),
because they themselves (by design) put burden on code-sharing, with other
reciprocal and non-reciprocal licenses. It seems to me, however, that
copy-left is an important pillar of the open-source community. Thus, the
aim of the exercise should NOT be solely to reduce burden on code-sharing.
Instead, I believe we ought to be producing maximal license compatibility
IN THE CONTEXT OF WHAT THE LICENSES AIM TO DO. That means not having
multiple licenses which do (or try to do) the same thing. There are some
pretty easy steps that can be taken here:
1. Non-templated licenses should not be tier-1
2. Perhaps we need some form of "do we trust these guys" before allowing
templated licenses with license maintainers in as tier-1.
3. Jurisdiction specific licenses should not be tier-1.
4. Licenses that have been deprecated by their submitter should not be
5. Licenses which have been superceded by a new version thereof should not
6. Licenses with well known legal flaws should not be tier-1 (I don't
think I'm being unfair if I put the Artistic license in this category
as IIRC its author has said as much).
7. If one or more licenses do not materially differ from eachother, then
all but those in extensive use shall not be tier 1. "Materially differ"
shall be considered both in terms of FUNCTION, and in terms of
substantive DRAFTING improvements.
I am guessing that if you went through the current set of approved
licenses, you'd strip out all but about 10 just on the above criteria.
That's not to say there shouldn't be other criteria, but I am guessing
the above is relatively uncontroversial (only 7 requires some subjectivity
on the part of the OSI).
The other thing we should be encouraging (as Larry points out) is reuse
of license terminology. If we had 10 license which were all drafted
in a similar manner, but "with or without the reciprocity clause" (etc.)
(think of the difference between OSL and AFL), then not only would we
minimize the time of people going through a license that is "new to them",
but case-law precedents (rare enough as they are) would be more widely
applicable, and we might just benefit from best-of-breed drafting.
More information about the License-discuss