A new package arrived on CRAN a few days ago. It offers a few headers files which wrap (parts) of the C API for R, but in a form that may be a little easier to use for C++ programmers. I have always liked how in Rcpp we offer good parts of the standalone R Math library in a namespace
R::. While working recently with a particular C routine (for checking non-ASCII characters that will be part of the next version of the dang package which collecting various goodies in one place), I realized there may be value in collecting a few more such wrappers. So I started a few simple ones starting from simple examples.
Currently we have five headers defines.h, globals.h, internals.h, math.h, and shield.h. The first four each correpond to an R header file of the same or similar name, and the last one brings a simple yet effective alternative to
UNPROTECT from Rcpp (in a slightly simplified way). None of the headers are “complete”, for internals.h in particular a lot more could be added (as I noticed today when experimenting with another source file that may be converted). All of the headers can be accessed with a simple
#include <tidyCpp> (which, following another C++ convention, does not have a
.hpp suffix). And a the package ships these headers, packages desiring to use them only need
As usage examples, we (right now) have four files in the
snippets/ directory of the package. Two of these, convolveExample.cpp and dimnamesExample.cpp both illustrate how one could change example code from Writing R Extensions. Then there are also a very simple defineExample.cpp and a shieldExample.cpp illustrating how much easier
Shield() is compared to
Finally, there is a nice vignette discussing the package motivation with two detailed side-by-side ‘before’ and ‘after’ examples that are the aforementioned convolution and
Over time, I expect to add more definitions and wrappers. Feedback would be welcome—it seems to hit a nerve already as it currently has more stars than commits even though (prior to this post) I had yet to tweet or blog about it. Please post comments and suggestions at the GitHub repo.