Brian J. Knaus

A colleague recently asked about getting started using Rcpp. After I sent the e-mail I thought that information may be of use to others. So I decided to blog about it.

Rcpp is a package on CRAN that facilitates an interface between R and C++. It includes data structures that allow for easy sharing of data between the two languages. Because writting C++ code is usually more challenging than writing R code you should probably reserve it for tasks that you will repeat. And if you’re going to repeat them you might as well put them in a package. I’ve put together my own version of a package skeleton that I use and you can find it here.

Once you have a package set up you’ll need to learn how to add code to it. Hadley Wickham has a nice introduction in his book on R packages and his book on advanced R. There is also the Rcpp Gallery. If you need some helpgetting started with C++ you might try their tutorial at cplusplus.com. And of course, there are our good friends Google and stackoverflow.

Compilers, oh yes, did I mention compilers? If you are working in a *nix environment (except OSX) you probably already have a compiler, usually gcc, so you should be good to go. If you are working on a Mac, you’ll need Xcode which includes clang. And if you’re working in Windows, you’ll need Rtools.

Once you’re up and running you might notice that there is something wrong with your code. Don’t feel bad, it happens to all of us! You’ll need to debug it. A simplistic way of dealing with this is to write to stdout at various points in your code.

Rcpp::Rcout << "We made it this far in the code!" << std::endl;

For something a little more sophisticated, Kevin Ushey has some nice posts on using debuggers with Rcpp such as valgrind and lldb, which is similar to gdb.

Those are the links I’ve used, and reused, to help myself learn Rcpp and C++. Have fun with compiled code in R!