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

Soeren Sonnenburg Soeren.Sonnenburg at first.fraunhofer.de
Tue Apr 8 05:08:48 EDT 2008


On Mon, 2008-04-07 at 14:02 +0000, Matthew Saltzman wrote:
> On Mon, 2008-04-07 at 09:58 +0200, Soeren Sonnenburg wrote:
> > On Sun, 2008-04-06 at 22:31 -0400, Matthew Saltzman wrote:
> 
> > 
> > The issue to me is that I and the majority of other opensource projects
> > cannot easily build upon CPL based work. The only workaround is that
> > projects that would want to use COIN-OR need to go through a license
> > change - which we all know is not so easy. 
> 
> Of course, this cuts both ways.

Yes you are right. Maybe, the only difference is that GPL based projects
are more widespread.

> > As a result coin-or is not as
> > widespread as it could be and its use is limited :( The best example is
> > the Open Solver Interface. It is part of COIN-OR and definitely
> > something people should use as it nicely interfaces to commercial as
> > well as free solvers. 
> > 
> > I would like to bring COIN-OR to debian, but my motivation to do so is
> > that I can use it in my software projects (e.g. shogun) and that it will
> > be useful for many other people too (scientific software is still rarely
> > used in debian from what I can tell - so extra limitations won't give us
> > more users, see e.g. http://qa.debian.org/popcon.php?package=glpk,
> > http://qa.debian.org/popcon.php?package=cvxopt or
> > http://qa.debian.org/popcon.php?package=lp-solve ).
> > 
> > > It's true that there are some restrictions we have to live with.  For
> > > example, we can't distribute binaries linked with GNU readline or GLPK.
> > > The onus is on the recipients of the code to build such binaries if they
> > > want those features.
> > 
> > No-one not just coin-or people can distribute binaries mixing GPL and
> > CPL that is the issue.
> 
> That's correct.  It doesn't serve much purpose to try to fix blame for
> this situation, but there's a symmetry at work here.  If you are a GPL
> fan, then the remedy for this state of affairs is for the other parties
> to change their licenses.  If you are GPL-averse, the solution is for
> GPL authors to make their code easier to link.

This is true. Nevertheless if a GPL project is big (many co-authors)
such a change can be close to impossible. With IBM being the sole
copyright holder of the coin-or core code it is at least only a single
entity that may decide...and if IBM makes the switch it is likely that
many of the projects will follow.
 
> > > The discussions on the COIN-OR board related to dual licensing are
> > > continuing, but given Alan's comment about clause 3.b, I'm not sure what
> > > progress we could make with IBM (who owns the code contributed by
> > > members of its staff).
> > 
> > I don't see the problem in 3.b:
> >
> > [license verbiage deleted...]
> 
> The original motivation for the CPL presumably had to do with
> dissatisfaction with GPL v2.  Given that the GPL v3 addresses the patent
> protection issue, it may be time to re-visit this question.  If the CPL
> "choice of law" clause is a narrow enough issue, it may be that dual
> licensing or a compatible CPL would be possible.
> 
> No breath-holding, though.  Moving lawyers (esp. at IBM) is not a quick
> or easy process.  I can make no promises except to try to revive the
> discussion.

I could not ask for more, so that is perfectly fine for me. I know that
it is a long term process, so I hope you find an advocate inside of IBM.
I will package CLP for debian in the meantime...

[...]
> > > > Please note that it is in any case not a good idea to use a GPL
> > > > incompatible license (see e.g.
> > > > http://www.dwheeler.com/essays/gpl-compatible.html)
> > > > After all most of the projects are GPL and a further incompatibility
> > > > between open source licenses does not really help...
> 
> I don't know of any scientific surveys to back up this conjecture, but I
> believe that some of the popularity of the GPL has to do with momentum
> (it's the most widely known, for many reasons).  A number of authors
> that I've spoken to know--and care--very little about the details of
> licenses and the consequences of their use, beyond a vague, general
> notion that they want to make their code available and perhaps that they
> want to avoid having it "taken private" without their knowledge or
> permission.  They pick the GPL because they've heard of it and they want
> to spend their time writing code rather than thinking about licenses. 

That is for sure one of the major reasons. Really, most people don't
care about licenses.

[choice of licenses]
> > Yes. Choice of license  really is about how one wants to see a program
> > used/distributed. BSD would potentially achieve the most widespread use
> > without necessarily getting contributions back, LGPL/MPL/CPL gets you
> > changes contributed back and well GPL ensures that any agglomerate work
> > will be open too...
> 
> Right.  Achieving a balance between widespread use and the FSF's freedom
> agenda is a delicate process.

Well it is not about the FSF here, it is about what you personally want.
(choose GPL if you want to make sure that it stays open forever).

> > The idea of open source is all great but with the license proliferation
> > and all the potential conflicts between licenses we loose many benefits
> > (there is an attempt to fix this, see
> > http://www.opensource.org/proliferation )
> 
> License proliferation is an issue for the FOSS community as a whole, but
> the community is made up of all kinds of people with different agendas.
> It's definitely bad for every project to create its own license with
> microscopic differences from other licenses just for vanity.  It's also
> bad when people with no legal training try to write licenses without
> carefully considering the consequences of their terms and wording.  But
> the idea of "one license to rule them all" is just impractical.

I totally agree.

On another subject, could one of the COIN-OR core people please add an
entry for coin-or at http://mloss.org ? mloss stands for machine
learning open source software and is a freshmeat like software index for
mloss. As many machine learners use optimization, it would be worth
making it known there too. 

Soeren
-- 
Soeren Sonnenburg - Fraunhofer FIRST      Tel: +49 (30) 6392 1882
Kekulestr. 7, 12489 Berlin, Germany       Fax: +49 (30) 6392 1805
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 189 bytes
Desc: This is a digitally signed message part
URL: <http://list.coin-or.org/pipermail/coin-discuss/attachments/20080408/39d6051d/attachment.sig>


More information about the Coin-discuss mailing list