[SciPy-dev] CeCILL license

[Gael Varoquaux]

On Sat, Oct 24, 2009 at 04:30:03PM -0500, Robert Kern wrote:
> None of the CeCILL licenses are suitable for code in scipy, in my
> opinion. The CeCILL-B is closest, but it has attribution requirements
> that are stronger in some ways than the GPL-incompatible advertising
> clause in the original BSD license. I'm pretty sure it is not
> GPL-compatible, and I doubt it will be considered DFSG-free or OSI
> approved.

I do believe it is DFSG-free. If not, I would love to know about an
example. According to the wikipedia page, CeCILL is GPL compatible, and
CeCILL-B is compatible with BSD, BIT and X11. However, the CeCILL webpage
does look suspicious. If you have any had facts, or previous legal or
informal cases involving the interpretation of the attribution closes, I
am interested, as I can try loby the legal office of my two employers,
which are 2 of the 3 CeCILL founder to make changes for the next version.

> While it is worthwhile for the author of the patch to ask Konrad, I'm
> pretty sure that code existed in ScientificPython before Konrad
> switched from the BSD license to the CeCILL license and could be
> forked from a previous version of ScientificPython.

+1 on that. I doubt that Konrad would mind relicensing a bit of code.

Cheers,

Gaël