Re: [sigc] Linking dynamically with SigC++ (now rather: License)

[Murray Cumming <murrayc murrayc com>]

On Thu, 2006-07-06 at 18:20 +0200, Aristid Breitkreuz wrote:
> Am Donnerstag, den 06.07.2006, 07:58 +0200 schrieb Murray Cumming:
> > > Am Mittwoch, den 05.07.2006, 18:49 +0200 schrieb Ulrich Eckhardt:
> > >> That's true. LGPL requires that people are able to relink your
> > >> executable with
> > >> a modified version of the LGPLed code, so either you ship objectfiles or
> > >> use
> > >> dynamic linking. Or, of course, provide the source.
> > >
> > > I was once told that sigc++ had no intent of being such restrictive. I
> > > asked for a more liberal license because such requirements are not
> > > acceptable for my needs. This is why for myself I switched to
> > > MPL/GPL/LGPL triple-license. But I was told that
> > >      1. sigc++ showed their intent clearly on the website.
> > >      2. changing license would be a long-term thing.
> > 
> > Yes, and nothing has changed since then:
> > http://mail.gnome.org/archives/libsigc-list/2006-February/msg00001.html
> 
> That was the mail I referred to. This is some relief but not sufficient.
> 
> > 
> > > But now I read that those restrictions DO apply.
> > 
> > You didn't read that from a maintainer. You will always hear different
> > opinions from different people on legal questions.
> 
> I do not think this is a legal question, rather it is a question of
> policy. And as such what matters is the policy of _each_ copyright
> holder. Obviously there are differences (Ulrich Eckhardts policy is
> stricter!?). Also it is not clear to me how binding a notice on a
> website can be.

Hence the need for an official exception.

> > As stated in that previous email, at some point we should explicit state
> > this in an exception in the headers, but I haven't got around to it, and
> > nobody has cared enough to write the exception text for us:
> 
> Or felt fluent enough in legalese for this. Or really understood the
> complications (I do not _fully_ comprehend them, too).
> 
> If I understand it correctly, the exception must be bullet-proof (of
> course ;-) ) and in difference to the LGPL allow the following
> use-cases:
>      1. Using all the template stuff (generally code in headers with
>         more than 10 lines per functional unit) in libsigc++ from
>         application / library code.

Yes.

>      2. Linking libsigc++ statically, at least on Windows or other
>         technically restricted platforms (I dislike Windows-DLLs).

Only a very small part of libsigc++ can be linked dynamically. I don't
think we will ever allow that part to be linked statically in
proprietary applications.

> Also do you want LGPL 2.1+ or LGPL 2.1 to be the base of the license
> (base+exception being the license).

At the moment, we say "either version 2.1 of the License, or (at your
option) any later version.", and I have no plans to change this. That
would be a separate and more difficult discussion.

> Another proposal would be tri-licensing MPL/GPL/LGPL as does Mozilla.
> This is the combination I use for my own free C++ library code. I
> basically hope that it's good enough.

I see no problem with LGPL+exception.

> Copyright and author's right (I will never again dare to mix those two)
> are complicated matters and of vast importance for software developers.

Still, nobody has cared enough yet to write that exception text.

-- 
Murray Cumming
murrayc murrayc com
www.murrayc.com
www.openismus.com