[SciPy-dev] [Fwd: Re: license status of your code on netlib]

Pauli Virtanen pav+sp@iki...
Fri Sep 11 09:02:38 CDT 2009


(Straying well inside the fields of OT.)

Fri, 11 Sep 2009 11:47:39 +0100, Robin wrote:
[clip]
> I guess I wondered what people thought about this. I'm looking for
> convincing arguments for people to stick to standard open source
> licenses where possible. I think the general impression from the above
> is that in most cases GPL would be more suitable than BSD - since they
> specifically want to prevent commercial exploitation.

GPL doesn't really prevent all commercial exploitation, as there may be 
no redistribution -- just someone using your code to do computations, in 
a for-profit company. Or used in a commercially supported GPLed 
application.

> But what do people suggest about the citation clauses and things like
> that?

This probably depends. If you trust people enough, you can just include a 
strong citation recommendation and how-to instructions, but not a license 
clause. I don't have experience on real-world consequences of either 
alternative.

OTOH, citations (in addition to the grapevine) are nearly the only metric 
that matters for academic software, so maybe you would like to safeguard 
this "income". It may be difficult and unproductive to argue against 
this. But I suppose the drawbacks might be that

(i) If you don't keep manually track of who have your software,
    and someone is rude enough to disregard a strong recommendation,
    they'd also just disregard the license. Or maybe not read it at all.

(ii) The advertisment requirement may annoy some users away, especially
     if your code is a component, not a complete analysis package.

(iii) They exclude the software from being "free software" in the sense
      some people (say, Debian) understand it. Stuff with such clauses
      won't land into Debian/Ubuntu main.

In a sense, also the drawbacks of the advertisement clause start to apply 
-- suppose many basic numerical codes had citation clauses. Most of the 
papers in all fields would have a citation to LAPACK. In a sense a good 
thing, in a sense a bit silly. Maybe a less of an annoyance than the 
advertisement clause, though, as you'd need to do this only once per 
paper.

-- 
Pauli Virtanen



More information about the Scipy-dev mailing list