Re: [eigen] Important: Relicensing Eigen to MPL2

[ Thread Index | Date Index | More Archives ]

On Mon, Jan 16, 2012 at 4:02 AM, Benoit Jacob <jacob.benoit.1@xxxxxxxxx> wrote:
> Hello,
> The CeCILL licenses are made by INRIA, which is Gael's employer. We've
> briefly looked at them, CeCILL-C is a somewhat decent LGPL
> alternative, but it's clear enough that it's less good than MPL2, that
> I don't a priori see the need to elaborate here. Let's just say:
> longer, less clear definitions/phrasing, tied to French juristiction,
> arguably less well-known internationally.

Cecill-C is incompatible with GPL, according to FSF, debian-science
and the local lawyer of my french uni.
So please avoid it. I have written to inria and they acknowledge what
some clarification could be made in next version.


> *** How is the MPL2 better? ***
> 1. The MPL2 is 2.8x shorter than the LGPL (including GPL, of which it
> is an addendum), see above table.
> 2. The MPL2 is much better written than any other copyleft-ish license
> I've seen. See for example these definitions from the MPL2:
>    1.6. “Executable Form”
>       means any form of the work other than Source Code Form.
>    1.13. “Source Code Form”
>       means the form of the work preferred for making modifications.
> Now if we compare to the GPL ( / LGPL), at first they look very similar:
>  The "source code" for a work means the preferred form of the work
> for making modifications to it.  "Object code" means any non-source
> form of a work.
> Except that they then proceed to talk about "executable works" without
> having defined _that_ term!
> 3. As a consequence of 2., the MPL is readily suitable for
> documentation (and examples) contrary to the LGPL / GPL. Indeed, its
> definitions of "source code" and "executable" forms are not specific
> at all to programs. And indeed, Mozilla has long been using the MPL
> for documentation and plans to do so with the MPL2 as well.
> 4. The MPL2 has a much simpler concept of copyleft. It's called
> file-level copyleft, and basically means that copyleft doesn't
> propagate across file boundaries. This should do wonders to ease
> concerns of people who worry about what they call "virality" (a term
> that I loathe).
> *** Why did we never consider the MPL1? ***
> For several reasons, the MPL1 wasn't a very compelling license for
> non-Mozilla projects. For example, it had a clause requiring legal
> disputes to be resolved under Californian jurisdiction...
> *** Should we relicense to MPL2-only or tri-license MPL2/LGPL/GPL? ***
> We should do MPL2 only. Mozilla has been tri-licensing MPL/LGPL/GPL
> for a while and that is considered a big mistake. It allowed people to
> screw us by forking mozilla code and releasing it under GPL only,
> preventing us from using it.
> Mozilla plans to switch from MPL/LGPL/GPL to MPL2 only, and the MPL2
> has been designed explicitly to allow that, so we know we're not in
> uncharted territory with this relicensing.
> *** But the MPL is not a well-known-enough license! ***
> Hard to tell: to some people it's very well-known, to some people it
> isn't. Recently, a large software company contacted us, spontaneously
> asking if we'd accept to relicence to MPL2, completely independently
> from our ongoing discussions! They weren't deterred by the recentness
> of the MPL2, saying that it was very close in spirit to the MPL1.
> I believe that the MPL2 is the only well-written, concise, no-nonsense
> copyleft license, so I expect it to gain a lot of traction within the
> copyleft world. Of course, copyleft != FOSS.
> Note that many questions about the MPL2 are answered in the MPL2 FAQ
> linked at the beginning of this email.
> Cheers,
> Benoit

Mail converted by MHonArc 2.6.19+