[R-pkg-devel] Licenses

[Marc Schwartz]

> On Oct 23, 2020, at 8:25 AM, Ege Rubak <rubak using math.aau.dk> wrote:
> 
> Hi all,
> 
> My two cents are below Marc's summary here:
> 
> On Thu, 2020-10-22 at 20:33 -0400, Marc Schwartz wrote:
>> Right now, the interpretation, without further clarification from
>> CRAN, would be, it is ok for a package to be on CRAN with license
>> based usage restrictions included (e.g. for non-commercial use), but
>> a package on CRAN, irrespective of it's own license, cannot
>> "interact" with other packages that do have restrictions...which
>> seems inconsistent.
> 
> It depends a bit what is meant by "interact". Years ago `spatstat` used
> `gpclib` with a non-commercial license to do polygonal operations. The
> solution was to list `gpclib` in `Suggests` and require the user to
> make an active choice to use this piece of software with a warning
> about non-commercial use. I find this to be an OK solution in lack of
> completely free alternatives. These days `gpclib` is still on CRAN and
> only has reverse `Suggests` and `Enhances`, so that seems fairly
> consistent.
> 
> In the long run this was unsatisfatory and our specific problem was
> solved by Adrian Baddeley by making the `polyclip` package.
> 
> Kind regards,
> Ege


Hi Ege,

The use of "Suggests" may be the relevant difference here. My use of the word "interact" was focused on the text in the CRAN policy that I quoted in my initial reply:

"Such packages are not permitted to require (e.g., by specifying in ‘Depends’, ‘Imports’ or ‘LinkingTo’ fields) directly or indirectly a package or external software which restricts users or usage."

As Duncan noted in his reply, there may be time based differences that are relevant here, if the CRAN policy had changed at some point, and there was, in effect, a grand-fathering of older packages that perhaps would not be accepted today under the current policy.

Regards,

Marc