1. Why would you extend R with C language?

• some parts of your program may run too slowly. One of the possible solutions is to rewrite them into C;

• if you create a library and you want it to be extremely fast, you will probably end up writing most of your functions in C.;

• it’s worth learning even the basic example, as most basic R functions are written in C.

A “Hello World” example

Let’s create a trivial C function, which sums elements of a given vector.

#include <R.h> 
#include <Rmath.h> 
#include <math.h>
#include <stdio.h>

void my_sum(double *x, int *n, double *s) {
    
    double sum = 0;

    for (int i=0; i<*n; i++) {
        sum = sum + x[i];
    }

    *s = sum;
}

As you can see:

• we do not provide a return clause; instead, we modify one of function’s arguments, or, being more precise, a place in computer’s memory at which this argument points;

• as every scalar is in fact a vector in R, the easiest way to pass elements from R to C is by using pointers.

Having the above function saved as my_sum.c, let’s compile it with R CMD SHLIB my_sum.c. Comparing to a standard compilation with gcc, ,R CMD SHLIB prodides additional information about special R libraries directories. Compilation will create two new files: my_sum.o and my_sum.so. How do we use them in R?

my_func_test.R

dyn.load("my_sum.so")

x <- c(1, 2, 3, 4)
sum_ <- .C("my_sum", x = x, n = length(x), s = 0)

sum_

## $x
## [1] 1 2 3 4
## 
## $n
## [1] 4
## 
## $s
## [1] 10

As you can see, the result is a list of function arguments, but some of their values (in this case s) were updated by our C function.

3. Usful links

As ‘C’ and ‘R’ are not the easient terms to search on the internet, here’s a short tutorial.