[R] licensing of R packages

[Prof Brian Ripley]

On Fri, 14 Nov 2008, Duncan Murdoch wrote:

> On 14/11/2008 4:42 AM, Carlos Ungil wrote:
>> I know the standard answer to this kind of question is "get legal
>> advice from a lawyer", but I would like to hear the (hopefully
>> informed) opinion of other people.
>> 
>> I would say that, according to the FSF's interpretation of the GPL,
>> any R code using GPL packages can be distributed legally only using
>> GPL-compatible licenses.
>
> I think they are talking about cases where the GPL libraries are compiled 
> into the new product.  Packages generally don't include copies of anything 
> from R, so our GPL doesn't apply to them.  (Writers may have chosen to copy 
> and modify base functions; if so, they are copying our code, and the GPL 
> would apply.)

That _has_ happened several times (usually without any credit and removing 
the copyright header from the R version), so package writers do need to be 
aware that base functions are not fair game.  Often it comes to light when 
the forked version fails when e.g. .Internal calls are changed.

I'm not going into the original question except to point out that R is 
licensed under GPL-2 and the quote was from the GPL-3 FAQ.  As FSF 
themselves insist, the two licences are incompatible.

-- 
Brian D. Ripley,                  ripley at stats.ox.ac.uk
Professor of Applied Statistics,  http://www.stats.ox.ac.uk/~ripley/
University of Oxford,             Tel:  +44 1865 272861 (self)
1 South Parks Road,                     +44 1865 272866 (PA)
Oxford OX1 3TG, UK                Fax:  +44 1865 272595