Re: License question

[Chris Vine <member cvine freeserve co uk>]

The LGPL has a specific exemption in section 5 for macros and "small inline functions" (I think a maximum length of 10 lines is specified to count as a "small inline function").  It does not include templates (whether of more or less than 10 lines) as it was not designed with C++ in mind.

I will look it up this evening and give you the precise text, and what I have done in the past as a modification of the licence (I have used a modified LGPL rather than a modified GPL, and if you wanted to I would have no problem with you adopting the same wording).
Chris


========================================
Message Received: Dec 13 2005, 10:25 AM
From: "Murray Cumming" 
To: "Chris Vine" 
Cc: gtkmm-list gnome org, doug dooglio net
Subject: *** SPAM *** Re: License question

On Mon, 2005-12-12 at 22:20 +0000, Chris Vine wrote:
> On Monday 12 December 2005 08:49, Murray Cumming wrote:
> > > The point is
> > > that if
> > > glibmm (not his code) contains templates released under an unmodified
> > > LGPL,
> > > he would as he says be required to release any source code which
> > > instantiates
> > > any of the templates or links (other than dynamically) with code which
> > > contains such instantiations. This would apply to anything using
> > > libsigc++
> > > (which means that although GTK+ can be used in closed source code, gtkmm
> > > cannot),
> >
> > [snip]
> >
> > This is highly debatable - otherwise nobody would be asking. The intention
> > is clear. If someone worries enough about this then they should ask the
> > FSF, who wrote the LGPL. In extreme circumstances, if it was really
> > necessary, we could relicense libsigc++ under the MIT/BSD license, or
> > license it as GPL+exception, as GNU's libstdc++ is licensed:
> > http://gcc.gnu.org/onlinedocs/libstdc++/17_intro/license.html
> >
> > Again, the only opinion I'd pay much attention to on this is the FSFs
> > because they have lawyers.
> 
> The FSF, who say they are the primary sponsor of GNU, recognise there is a 
> problem as libstdc++ is released under a modified GPL to deal specifically 
> with the template problem, as you yourself note. The problem with the LGPL 
> is explicitly set out in the web page to which you refer, so they at any rate 
> do not regard it as "highly debatable".

Still, I'd prefer to hear directly from an FSF person about this because
it does seem vague. I'd like to know
a) Is LGPL meaningless for C++ libraries that provide templated types
(most C++ libraries)?
b) Is LGPL meaningless for C libraries that have macros in their
headers?
c) Is a certain amount of a) or b) OK?

[snip]

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