Re: freeness and compatibility of CeCILL-C licence
On 30/03/2017 03:27, Drew Parsons wrote: On Thu, 2017-03-23 at 10:51 +0800, Drew Parsons wrote: If I'm reading that right, we can link it from BSD and LGPL libraries. Currently MUMPS is in Debian used by getfem++ LGPL petsc BSD-2 which is used by dolfin LGPL trilinos BSD code-aster GPL2 So there should be no conflict using mumps with CeCILL-C, except perhaps with code-aster. Graham Inggs wrote to me to say: Aster in Debian currently FTBFS and is unmaintained. It is probably a good candidate for removal. Aster has been removed from Debian unstable, see https://packages.qa.debian.org/a/aster/news/20170404T193602Z.html
Re: freeness and compatibility of CeCILL-C licence
On Thu, 2017-03-23 at 10:51 +0800, Drew Parsons wrote: > > If I'm reading that right, we can link it from BSD and LGPL > libraries. > Currently MUMPS is in Debian used by > > getfem++ LGPL > petsc BSD-2 > which is used by dolfin LGPL > trilinos BSD > code-aster GPL2 > > So there should be no conflict using mumps with CeCILL-C, except > perhaps with code-aster. Graham Inggs wrote to me to say: Aster in Debian currently FTBFS and is unmaintained. It is probably a good candidate for removal. Drew
Re: freeness and compatibility of CeCILL-C licence
On Thu, 23 Mar 2017 10:38:37 + Ian Jackson wrote: > Dmitry Alexandrov writes ("Re: freeness and compatibility of CeCILL-C > licence"): > > [Ian:] > > > (IMO it would not be fine if it specified Russian or Chinese courts.) > > > > Interesting idea. Any substantiation for such a discrimination of origin? > > Some courts are more trustworthy than others. Personally I don't agree with this discrimination: I think that the normal venue selection rules should never be subverted by Free Software licenses, regardless of how palatable may look the potential pre-selected venue... -- http://www.inventati.org/frx/ There's not a second to spare! To the laboratory! . Francesco Poli . GnuPG key fpr == CA01 1147 9CD2 EFDF FB82 3925 3E1C 27E1 1F69 BFFE pgpJtikLloSYo.pgp Description: PGP signature
Re: freeness and compatibility of CeCILL-C licence
On Thu, 23 Mar 2017 02:47:43 +0300 Dmitry Alexandrov wrote: [...] > > Francesco Poli dislikes the choice of law and courts clause, but I > > think it's fine. For the record, I think a choice of law clause is acceptable. On the other hand, I consider a choice of venue clause as a non-free restriction. > > IBM PL v1.0 contains a choice of law clause and it’s listed as > suitable for Debian’s main [...]. It seems to me that the IBM PL v1.0 only includes a choice of law clause and no choice of venue. But discussing that license would take us far away from the issue at hand... > > As for arbitration clause, could anyone explain, what’s the practical > difference between ‘choice of law of N’ and stating that disputes > should be resolved in *general jurisdiction* courts of N? > IMHO, they are effectively the same. As far as I can tell, a choice of law clause only states that any dispute must be litigated following the laws of a pre-selected jurisdiction. On the other hand, a choice of venue clause insists that any dispute be litigated in the courts of a pre-selected country (or even state, region, local area, ...). If a user is sued (say, for copyright infringement) by the copyright holder, that user will be forced to travel possibly long distances in order to defend him/herself. -- http://www.inventati.org/frx/ There's not a second to spare! To the laboratory! . Francesco Poli . GnuPG key fpr == CA01 1147 9CD2 EFDF FB82 3925 3E1C 27E1 1F69 BFFE pgpZ3lyqW5AZk.pgp Description: PGP signature
Re: freeness and compatibility of CeCILL-C licence
On Thu, 23 Mar 2017 11:04:48 + Ian Jackson wrote: > Drew Parsons writes ("Re: freeness and compatibility of CeCILL-C licence"): > > If I'm reading that right, we can link it from BSD and LGPL libraries. > > Currently MUMPS is in Debian used by > ... > > code-aster GPL2 > > This is a problem then. Although we seem to disagree on the DFSG-freeness of the CeCILL-C license, I am happy to see that you agree with me on its GPL-incompatibility and on the fact that GPL-licensed programs or libraries linked with CeCILL-C-licensed libraries are indeed a problem. I reported a number of such bugs [1] years ago and tried multiple times to solve the issue for SCOTCH (which has already been mentioned in this same thread: it's another library released under the terms of the CeCILL-C license and often linked with GPL-licensed code). [1] https://bugs.debian.org/cgi-bin/pkgreport.cgi?archive=both;tag=scotch-license-issues;users=debian-science-maintain...@lists.alioth.debian.org Unfortunately, the maintainers of the affected Debian packages do not believe me and want to see an official statement from the Debian FTP Masters. Such official statement has been requested multiple times for years, but never seems to arrive. I contacted the author of SCOTCH more than once, but failed to persuade him that SCOTCH should be re-licensed (or dual-licensed) under the GNU LGPL v2.1; he seems to be no longer willing to reply to my e-mail messages. > > Is there any possibility of CeCILL being persuaded to add a GPL > compatibility exception ? The licence already has an upgrade clause > IIRC. Maybe this is the way to go, thanks for suggesting this strategy! Because of the license version upgrade clause, it would solve many issues at the same time (not only for SCOTCH and MUMPS, but also for other CeCILL-C-licensed works). Maybe we should contact <le...@cecill.info> [2]. [2] http://www.cecill.info/contacts.en.html Are you willing to help? Please let me know, thanks for your time and for any contribution you may provide. -- http://www.inventati.org/frx/ There's not a second to spare! To the laboratory! . Francesco Poli . GnuPG key fpr == CA01 1147 9CD2 EFDF FB82 3925 3E1C 27E1 1F69 BFFE pgpQ0P69y1Yxn.pgp Description: PGP signature
Re: freeness and compatibility of CeCILL-C licence
On Thu, 2017-03-23 at 11:04 +, Ian Jackson wrote: > Drew Parsons writes ("Re: freeness and compatibility of CeCILL-C > licence"): > > If I'm reading that right, we can link it from BSD and LGPL > > libraries. > > Currently MUMPS is in Debian used by > > ... > > code-aster GPL2 > > This is a problem then. > > Is there any possibility of CeCILL being persuaded to add a GPL > compatibility exception ? The licence already has an upgrade clause > IIRC. In a sense it's a problem that code_aster needs to push to get fixed. code_aster also comes from France, they could sort it out face to face. My guess is they're both using the licences perfunctorily, with MUMPS contractually obliged by their workplace to use a CeCILL. Scientific programmers (at least, the kind that build free software) tend to roll with the general spirit of free software without particular care for the legal subtleties. That is, as far as the authors are concerned, there is no violation (this is my interpretation or guess of what they're thinking). If the other library you're using is in a separate file and only accessed at runtime, then why should its free licence affect the free licence of your program? It's absurd to think that linking to libraries matters to your licence (assuming their licence doesn't actually forbid linking, in which case it wouldn't particularly be a free licence). Of course the question of linking is one of the reasons for changing GPL2 to GPL3. Maybe we need some billionaire friend of free software to send the matter to court, fund both sides of the argument and have it settled once and for all. In any case, I've asked the MUMPS developers to read this thread and let us know what they think. Drew
Re: freeness and compatibility of CeCILL-C licence
Drew Parsons writes ("Re: freeness and compatibility of CeCILL-C licence"): > If I'm reading that right, we can link it from BSD and LGPL libraries. > Currently MUMPS is in Debian used by ... > code-aster GPL2 This is a problem then. Is there any possibility of CeCILL being persuaded to add a GPL compatibility exception ? The licence already has an upgrade clause IIRC. Ian.
Re: freeness and compatibility of CeCILL-C licence
Dmitry Alexandrov writes ("Re: freeness and compatibility of CeCILL-C licence"): > [Ian:] > > (IMO it would not be fine if it specified Russian or Chinese courts.) > > Interesting idea. Any substantiation for such a discrimination of origin? Some courts are more trustworthy than others. Ian.
Re: freeness and compatibility of CeCILL-C licence
On Wed, 2017-03-22 at 13:07 +, Ian Jackson wrote: > Drew Parsons writes ("freeness and compatibility of CeCILL-C > licence"): > > There are various discussions about the status of the CeCILL-C > > licence > > v1 (and other CeCILL licences) in the history of this mailing > > list. > > It's not listed at https://www.debian.org/legal/licenses/ > > but when it last came up on this list, Thibaut Paumard suggested > > it's > > fine, LGPL compatible, > > https://lists.debian.org/debian-legal/2010/01/msg00064.html > > Is this still the consensus? > > ... > > CeCILL-C v1 itself is http://www.cecill.info/licences/Licence_CeCIL > > L-C_V1-en.html > > I think this is a DFSG-free GPL-incompatible copyleft licence. > > It's GPL-incompatible because it is not identical to the GPL and > requires derivatives to have the same licence. If I'm reading that right, we can link it from BSD and LGPL libraries. Currently MUMPS is in Debian used by getfem++ LGPL petsc BSD-2 which is used by dolfin LGPL trilinos BSD code-aster GPL2 So there should be no conflict using mumps with CeCILL-C, except perhaps with code-aster. Incidentally, SCOTCH is also CeCILL-C, and is already in the archive (and also used by many of these packages). > Francesco Poli dislikes the choice of law and courts clause, but I > think it's fine. I suppose it is a bit of an annoying clause. Then again, if a litigant wants to fly me to Paris to testify, maybe I can live with that... > (IMO it would not be fine if it specified Russian > or > Chinese courts.) Bit extreme to declare entire nation's jurisprudence corrupt. Unless you mean the court's judgement will be ignored, in which case it doesn't matter where the court is anyway.
Re: freeness and compatibility of CeCILL-C licence
[Sorry for sending unfinished letter.] > Francesco Poli dislikes the choice of law and courts clause, but I > think it's fine. IBM PL v1.0 contains a choice of law clause and it’s listed as suitable for Debian’s main [0]. As for arbitration clause, could anyone explain, what’s the practical difference between ‘choice of law of N’ and stating that disputes should be resolved in *general jurisdiction* courts of N? IMHO, they are effectively the same. [0] https://www.debian.org/legal/licenses/ > (IMO it would not be fine if it specified Russian or Chinese courts.) Interesting idea. Any substantiation for such a discrimination of origin?
Re: freeness and compatibility of CeCILL-C licence
> Francesco Poli dislikes the choice of law and courts clause, but I > think it's fine. (IMO it would not be fine if it specified Russian or > Chinese courts.) Interesting idea. Any substationation for such a discrimination of origin?
Re: freeness and compatibility of CeCILL-C licence
Drew Parsons writes ("freeness and compatibility of CeCILL-C licence"): > There are various discussions about the status of the CeCILL-C licence > v1 (and other CeCILL licences) in the history of this mailing list. > It's not listed at https://www.debian.org/legal/licenses/ > but when it last came up on this list, Thibaut Paumard suggested it's > fine, LGPL compatible, > https://lists.debian.org/debian-legal/2010/01/msg00064.html > Is this still the consensus? ... > CeCILL-C v1 itself is > http://www.cecill.info/licences/Licence_CeCILL-C_V1-en.html I think this is a DFSG-free GPL-incompatible copyleft licence. Francesco Poli dislikes the choice of law and courts clause, but I think it's fine. (IMO it would not be fine if it specified Russian or Chinese courts.) It's GPL-incompatible because it is not identical to the GPL and requires derivatives to have the same licence. The warranty disclaimer clauses are unusually worded but seem to protect contributors (including third parties) well enough. Ian.
Re: freeness and compatibility of CeCILL-C licence
On Tue, 21 Mar 2017 23:51:12 +0800 Drew Parsons wrote: > There are various discussions about the status of the CeCILL-C licence > v1 (and other CeCILL licences) in the history of this mailing list. The CeCILL-C license is a GPL-incompatible license, which may even be considered to fail to meet the DFSG. I personally consider works (solely) released under the terms of the CeCILL-C license as non-free. See https://lists.debian.org/debian-legal/2008/01/msg00171.html for an analysis of the license. It's GPL-incompatible, since it includes some restrictions not included in the GPL (at the very least, the choice of venue clause) and has no explicit conversion-to-GPL clause. Perhaps it's LGPL-compatible, but that's not especially interesting, since being LGPL-compatible is not difficult... > It's not listed at https://www.debian.org/legal/licenses/ > but when it last came up on this list, Thibaut Paumard suggested it's > fine, LGPL compatible, > https://lists.debian.org/debian-legal/2010/01/msg00064.html > Is this still the consensus? Thibaut just stated that it's LGPL-compatible. I cannot find any statement about it being "fine". > > The French scientific computing community have been been quite active, > with many packages released under CeCILL licences. Indeed, and they increased license proliferation, which is really really bad... :-( > If CeCILL-C is an acceptable licence, [...] I personally do not consider it acceptable. Please note that I am a Debian Project external contributor (I am not a member of the Project) and I do not speak on behalf the Debian Project. What I expressed are my own personal opinions. -- http://www.inventati.org/frx/ There's not a second to spare! To the laboratory! . Francesco Poli . GnuPG key fpr == CA01 1147 9CD2 EFDF FB82 3925 3E1C 27E1 1F69 BFFE pgpYbj0ql3ua2.pgp Description: PGP signature