[SCILAB] License of routine rpoly.f

[Olivier Robert]

Hello Sylvestre

> Do you know a free implementation of this algorithm ?

I found one used in a scientific software developped at the CERN.
http://geant4.cern.ch
I didn't tried the code, but it implements the same algorithm as TOM
493. (ie. the Jenkins-Traub algorithm)
http://geant4.cern.ch/bin/SRM/G4GenDoc.exe.pl?flag=2&FileName=G4JTPolynomialSolver.hh&FileDir=source/global/HEPNumerics/include
It is an open source software with a non standard licence.
http://geant4.web.cern.ch/geant4/license/LICENSE.html
At first glance it looks like the GPL but with restrictions about
patents. What do you think?

The GNU Scientific Library also includes a root finding routine but it
doesn't implements Jenkins-Traub.

Regarding the licence of the fortran routine, maybe there is still
the possibility to ask the permission to the ACM to deliver the code as
GPL.

Best regards,

Olivier Robert

On Thu, 04 Feb 2010 17:13:50 +0100
Sylvestre Ledru <sylvestre at debian.org> wrote:

> Hello Olivier,
> 
> Le lundi 01 février 2010 à 14:49 +0100, Olivier Robert a écrit :
> > Hello,
> > I'm contacting the debian maintainers of the Scilab package to get
> some
> > clarifications about the license of a particular routine used in
> Scilab.
> > 
> > ./scilab-5.1.1/modules/polynomials/src/fortran/rpoly.f
> > 
> > According to the header included into the file, the code is
> > copyrighted by the INRIA and placed under the CeCILL license.
> > It is also stated that the code is from NETLIB : TOMS/493. In fact
> > it is a modified version of the original code available from
> > http://www.netlib.org/toms/493.
> Indeed. It was a mistake, I forwarded that upstream and Vincent
> Couvert removed 
> this issue
> ( http://gitweb.scilab.org/?p=scilab;a=commitdiff;h=298062c4634227b92f44910b3d47552b023cb292 ).
> 
> 
> > According to http://www.netlib.org/toms/ the "use of ACM Algorithms
> > is subject to the ACM Software Copyright and License Agreement".
> > Further information is given on the page
> > http://www.acm.org/publications/policies/softwarecrnotice/
> > The licence grants the right to execute, copy, modify and distribute
> > the code and the binary only for non-commercial use. But for
> > **commercial use, you have to get the authorisation from the
> authors.**
> > I'm quite confused by this situation. What licence should I read ?
> Well, this ACM license issue is still unclear. Old ACM algorithms were
> in a kind of public domain but they updated their license policies and
> try to make them retroactive.
> But anyway, yes, there is an issue here. Do you know a free
> implementation of this algorithm ?
> 
> Thanks for pointing this out,
> Sylvestre
>