[Coin-discuss] COIN-OR licences again...

[Matthew Saltzman]

On Fri, 2008-04-11 at 15:00 -0400, kevin.c.furman at exxonmobil.com wrote: 
> >> As for the dual licensing idea, that has been discussed on and off for a
> 
> >> long time and there is very little chance of it happening. However, we
> 
> > Could you give a reason why this is unlikely? That it has been discussed
> > a couple of times just underlines that there is a need to change
> > things...
> 
> ...
> 
> > Well that does not really sound dispassionate. If the reason why dual
> > licensing is not an option is that there is a strong dislike ?against
> > any GPL compatible license from the people in charge here, then nothing
> > will change. If the aim is to see coin-or projects widely used however
> > it is important to consider dual licensing with another more compatible
> > license. As I guess the goal for IBM to open sourcing this project was
> > so see it widely used and extended I would hope that a solution can be
> > found.
> 
> I don't believe there is any particular dislike of GPL compatible licenses.
> I think the main concern is that IBM chose the wording of the CPL for a
> particular reason even though it was known to be incompatible with the GPL.
> Many of the contributors to COIN-OR are IBM employees, and thus are not the
> decision makers when it comes to intellectual property derived from their
> efforts.  The primary people needing convincing on dual licensing with a
> GPL compatible licene are not COIN-OR contributors, but Big Blue itself.
> Although IBM's primary goal may be to see the projects widely used, they
> probably also have other concerns related to protecting their patent
> portfolio and control of the actual CPL terms which makes it unlikely they
> would ever allow for the dual licensing any of the code under GPL or LGPL.
> 
> License is a choice.  A choice was made.  Regardless of how much easier it
> might make life for others, I think we need to respect someone's choice in
> how they want their intellectual property licensed.
> 
> In my opinion, of one wanted to pursue this line of thought, the path of
> least resistance would be to look at the CPL and analyze exactly what terms
> would need to be modified at a minimum to make it GPL compatible.  From
> that point then try encourage creation of a new version of the CPL such
> that current CPL code could all be licensed under the new version.

The main reason the FSF thinks the CPL is incompatible with GPLv3 is the
"choice-of-law" clause[1].  I believe that's the part of the CPL that
says that disputes will be adjudicated in New York.  If that's the only
issue, then we need to find out if that clause could be relaxed in the
CPL or if IBM would consider licensing under GPLv3 or LGPLv3 (without
the choice-of-law clause) what they would not release under GPLv2 or
LGPLv2.

I'd guess that LGPL would be the GNU license of choice for us, as (like
the CPL) it does not transfer across API boundaries.

[1] That's different than the GPLv2, which differed in not including
patent clauses.
-- 
                Matthew Saltzman

Clemson University Math Sciences
mjs AT clemson DOT edu
http://www.math.clemson.edu/~mjs