Bug#618696: closed by Adam C Powell IV <hazelsct at debian.org> (Re: elmer: multiple licensing issues)

Adam C Powell IV hazelsct at debian.org
Wed Apr 13 21:27:29 UTC 2011


On Wed, 2011-04-13 at 23:00 +0200, Francesco Poli wrote:
> On Wed, 13 Apr 2011 16:43:37 -0400 Adam C Powell IV wrote:
> 
> [...]
> > Hi Francesco,
> 
> Hi Adam!
> 
> > 
> > You mentioned elsewhere (I think your bug against salome) that the
> > CeCILL-C license of scotch is not GPL-compatible.  (Can you describe why
> > or send a link?)
> 
> Sure, here's an analysis of the CeCILL-C license:
> http://lists.debian.org/debian-legal/2008/01/msg00171.html

Very interesting, but also disappointing.  Thanks for the link.

> > If that's the case, then Elmer has a problem because it links with
> > Scotch.
> 
> Well spotted, I didn't notice it!
> 
> > That's not one of the original issues of this bug, but is very
> > much a licensing issue.
> 
> It definitely seems to be one.
> 
> > 
> > More seriously, upstream distributes Elmer with METIS, which is very
> > non-free, and with no linking exception.  (The .dfsg package removes the
> > METIS code from the tree.)  As the copyright holders, this is their
> > prerogative, but anyone else who distributes them together risks a
> > copyright violation.
> > 
> > Let me see if upstream will work with me on this, as they did a linking
> > exception for OpenCASCADE.
> 
> I think the possible solutions, in descending order of desirability,
> are:
> 
>  (A) SCOTCH copyright holders should be contacted and persuaded to
> re-license (or dual-license) it under GPLv2-compatible terms.
> 
>  (B) SCOTCH should be substituted with a GPLv2-compatible
> replacement, if any is available.
> 
>  (C) GPL-licensed work (Elmer and any other work that indirectly links
> with SCOTCH through Elmer) copyright holders should be asked to add
> license exceptions that give permission to link their work with code
> released under CeCILL-C v1.0 .

I agree with this order.  But B is very unlikely, I have been looking at
this type of software, and it has taken several years for François to
bring Scotch close to the capabilities of (Par)METIS.  There are not
many graph theorists/programmers in the world who can do this, and it
takes a very long time.

> Before contacting Elmer upstream (and possibly other GPL-licensed work
> upstream), I would try solution (A), or maybe (B).
> Adding linking exception should be regarded as a last resort strategy...

Okay, I'll try A, though I've already started on C.  We'll see.

-Adam
-- 
GPG fingerprint: D54D 1AEE B11C CE9B A02B  C5DD 526F 01E8 564E E4B6

Engineering consulting with open source tools
http://www.opennovation.com/
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 190 bytes
Desc: This is a digitally signed message part
URL: <http://lists.alioth.debian.org/pipermail/debian-science-maintainers/attachments/20110413/264ee89c/attachment.pgp>


More information about the debian-science-maintainers mailing list