[Rd] R-base licensing question

[Logan Lewis]

On Sunday 17 September 2006 1:53 am, you wrote:

> I'm not sure what you are asking, but in general R's GPL license is
> completely irrelevant unless you are distributing R.  If you're
> writing a package and distributing only your own work, you can
> license it as you like.
>[snip]

I realize how semantic this issue seems, but it boils down to what 
constitutes a derivative work under copyright.  The standard (at least 
as the FSF sees it) is not whether you include GPL code for 
distribution, but whether you link to GPL code at all (even 
dynamically).  This is, as you note, why the LGPL exists.

My point is that interpreted languages blur the lines (I've found 
similar discussions for other programming languages).  Most of the 
time, the interpreter is clearly distinct from software written in the 
interpreted language.  Perl is an example.  Write Perl code, using the 
base language, and you can license it however you'd like.  However, 
this particular FAQ argues that if you _use_ (not even include) GPL 
code written in Perl (like a Perl module), then you must distribute 
your code with a GPL-compatible license.  It certainly struck me as a 
strict interpretation, but when you compare it to linking to a 
non-standard library of a compiled language, it is analogous.  This is 
why the standard libraries in GCC are LGPL or have exceptions for 
non-GPL linking.  

What makes R relatively unique is that almost everything a user will 
code calls functions from R-base, written in R itself (i.e. just 
another package, really).  Certainly other R packages would be in the 
same position as the Perl modules mentioned as an example from this 
FAQ.  That is, if you use a package's functionality in your code, you 
must distribute your code in accordance with the terms of the package's 
license. 

Regards,
Logan