Re: Recommendation for a CL data structures library

[Raffael Cavallaro]

On 2010-03-25 09:59:52 -0400, Tamas K Papp said:

> I disagree -- I don't think that the FSF considers the GPL a "poor
> fit" for libraries.  Quite the opposite (see [1]).  They just
> recognized that in certain situations, some people would prefer
> something like the LGPL, and I guess that they wanted to give them the
> choice.  But the GPL is still the option they recommend, even for
> libraries.
>
> [1] http://www.gnu.org/licenses/why-not-lgpl.html

I don't put much stock in such rationalizations. IOW, having released 
the Library GPL, they realized that it was in many ways superior to the 
GPL from both the user and the open source perspective. They've been 
backpedaling ever since, and the frankly silly renaming of the Library 
GPL to the Lesser GPL is a clear sign of this ongoing attempt at damage 
control.

I.e., when I say they recognize that the GPL is a poor fit for 
libraries, I'm saying that their actions (release of the LGPL and 
subsequent renaming) speak louder and more convincingly than their 
words (the link you provide).

> Regarding the broader issue (of how people license their libraries): I
> think this is an optimization problem where people have heterogeneous
> objective functions, and thus trying to convince people to pick
> another license is not always a worthwhile.
>
> It is possible that someone using a GPL/LGPL/LLGPL/BSD/MIT/... license
> is perfectly aware of the advantages and disadvantages, it is just that
> they decided to make a different choice.  In which case, threads like these
> are unlikely to be fruitful.

I don't think their objective functions differ much from mine. I think 
they don't appreciate how the license plays out in the real world. 
Those who support the GPL for libraries think that by doing so they 
maximize the promotion of open source. I contend that the LGPL or 
Apache or APSL license lead to greater amounts of open source because a 
GPL library excludes one of the largest pools of possible contributors 
- professional developers who work on closed source projects. These 
potential contributors will instead either

1. reinvent that particular wheel in a closed source fashion (loss to 
free software)
2. use a library with a license that doesn't require any publication 
such as the bsd, or mit. (possible loss to free software)
3. use a library with a license that requires publication only of 
covered code such as the LGPL, APSL, Apache, etc. Only this last case 
inevitably results in more open source.

So by releasing a library under the GPL one provides as many ways for 
open source to lose as to win. Choosing the winning path in the first 
place by releasing the library under the LGPL/LLGPL/Apache etc. license 
leads to the biggest gains for open source. Again, the recognition of 
this reality is what led to the Library GPL in the first place. So 
people who support the GPL for libraries are unwittingly advocating for 
freedom in a way that actually results in less open source.

Even if I don't convince my correspondents here, I do hope that some of 
those reading this thread will develop a more nuanced view of open 
source licenses. I've said what I have to say, so (undoubtedly much to 
your relief), I'll stop.

warmest regards,

Ralph
--
Raffael Cavallaro