[SciPy-User] ANN: graph-tool, a package for graph analysis and manipulation

Gael Varoquaux gael.varoquaux@normalesup....
Thu Dec 3 16:38:53 CST 2009

On Thu, Dec 03, 2009 at 11:23:58PM +0100, Tiago de Paula Peixoto wrote:
> The issue of which is the best type of license aside, do you think that
> switching to LGPL would make it less "useless" to the SciPy community,
> or does it need to be absolutely non-copyleft, BSD-like?

>From a practical point of view, I can use a LGPL-licensed library in much
more situations than I can use a GPL-licensed library, because it does
impose restrictions onto the users of my code. As a result, if my
research yields practical application, I can still 'sell' the code.

Don't worry, the code I am talking about is research-grade code that gets
pushed to BSD-licensed libraries when it is not so ugly that I can't even
show it. However, I have found that having the argument that one day, I
may be able to sell that code (I would be selling my expertise more than
the code) helps a lot convincing management that we can release the code
under an open source license. Most researchers don't even release code,
which means that the restrictions of the GPL don't apply, but also that
scientific collaboration and the software ecosystem both pay a hefty

So yes, LGPL helps a lot.

And please, don't take the comments as pressure (even though they
probably are, unwillingly), or an ideological position. As Robert pointed
out, you own your work, and we should all be grateful to you for sharing
it. I am just giving the rational behind my position not to use copyleft
licenses for my own work (if you are interest in the LGPL vs BSD
discussion, feel free to ask, but it seems to me off topic with regards
to your question).



More information about the SciPy-User mailing list