Re: Gtkmm, LGPL and C++

[Neil <mail neilramsden co uk>]

Chris, Jonner

Thanks for the pointers - and confirmation of what Gtkmm intends.

My issue is with templates rather than sub-classing. There's a fair bit 
on the latter that I also found that suggests sub-classing needn't be an 
issue.

If LGPL is loose enough to account for templates, I don't understand why 
GNU libstdc++ needs a runtime exception. Both Gtkmm and libstdc++ have 
similar constraints and intents. The only difference is that templates 
probably play a more pivotal role in libstdc++, but they do also have a 
necessary one in Gtkmm (especially via sigc++).

Eg. see:
http://gcc.gnu.org/onlinedocs/libstdc++/17_intro/license.html

The new, draft LGPL v3 does mention templates of "ten or fewer lines", 
so maybe the explicit language is getting there. I guess the rationale 
is that they're roughly equivalent (for licensing) to a macro.

There are also existing alternatives like GPL+runtime exception which 
seem to give you what you intend, though I might have missed something here.

It'd be interesting to see how well the LGPL's "ten or fewer lines" 
condition works in actual libraries (eg. signal.h in sigc++-2.0 has some 
hunkier function templates); or how meaningful it is when one template 
could call another.

On this particular condition, I can see that the LGPL may want to limit 
the effect of header code in some way, but "ten or fewer lines" seems 
inappropriate and distinctly uninviting for most of us who don't want to 
count function lengths before we know what licensing restrictions there 
are on a piece of software. In practice, macros, templates and inline 
code in headers will be self-limiting, so the condition might also be 
redundant.

I did try the archive, but must've used poor search terms (LGPL, 
proprietary, commercial, etc) or looked in the wrong place. Any 
suggestions here welcomed.

Thanks again,
Neil.

Chris Vine wrote:
> On Tuesday 20 February 2007 17:44, Jonathon Jongsma wrote:
>> I believe there was some talk in the past about adding explicit
>> language to the license regarding templates, but nobody has cared
>> enough to propose language so far.
>
> That is not correct.  Look at the archive.
>
> Chris

Jonathon Jongsma wrote:
> This is a question that pops up on the list every couple weeks or
> months.  The last time somebody asked, murray said[1]:
> "There is no question that gtkmm may be used by closed-source
> proprietary applications."
>
> I believe there was some talk in the past about adding explicit
> language to the license regarding templates, but nobody has cared
> enough to propose language so far.  If you feel you need additional
> language in the license, maybe you could propose something?  I think
> the intention of the developers is clear, but if you don't feel
> comfortable with that, and you're willing to work with us, we might be
> able to clear up the ambiguity.
>
> Of course, Murray's opinion would be the one to really pay attention
> to, since he's done most of the work and I'm just a minor contributor.
>
> [1] http://www.mail-archive.com/gtkmm-list gnome org/msg07058.html