[R-pkg-devel] Including Open-Source C Code in R Package

[Duncan Murdoch]

On 05/08/2015 4:14 PM, Zhu, Zijie wrote:
> Hi Dirk and Duncan,
> 
> Many thanks for replying! I will definitely try drat and see how it works.
> 
>> I don't think so --- how could anyone use your package, if it is incomplete?
>>
>> You should get permission from the authors of the other package to
>> release it under one of the allowed licenses.  Or distribute the whole
>> thing on Github, or some other non-CRAN location.
> 
> And sorry, Duncan, that I didn't explain my situation very clearly.
> I'm also co-author of the other R package. At the beginning, I
> co-authored an R package that contains the open-source C model. I
> tried to submit this package to CRAN, but the CRAN maintainer says the
> license of that C model is not in mainstream and I should obtain a
> written confirmation (which I failed, cuz the institution that wrote
> the C code never responded.) So I tried to split the package into
> halves, one half containing the C-model code and the other half
> without C code. I intend to put the half without C code on Github,
> while submitting the half with C code to CRAN. So I'm actually
> co-author of both R packages that I referred to.
> 
> The .onLoad funciton above is trying to download the R package with C
> code from Github, when users first install the R package without C
> code from CRAN. So when user type install.packages("CRAN-part"), she
> will also (unknowingly) call install_github("Github-part"). I am
> concerned that CRAN maintainer will not be happy about this .onLoad
> function and what it does.
> 

So don't submit it to CRAN.  People use CRAN because they have
standards, and it sounds as though you are trying to find a way to
subvert those standards.

Duncan Murdoch