75 lines
2.6 KiB
Plaintext
75 lines
2.6 KiB
Plaintext
|
Rcpp version 0.7.0 went onto CRAN this weekend. The key new features are
|
||
|
|
|
||
|
|
o inline support, taken from Oleg Sklyar's neat inline package and
|
||
|
|
adapted/extented to support Rcpp as well as external libraries
|
||
|
|
(see below for an example); this even works on Windows (provided you
|
||
|
|
have Rtools installed and configured);
|
||
|
|
|
||
|
|
o addition of a new simple type RcppSexp for importing or exporting
|
||
|
|
simple types directly between R and C++
|
||
|
|
|
||
|
|
o addition of a number of new examples for both these features
|
||
|
|
|
||
|
|
o code reorganisation: every class has in its own header and source file
|
||
|
|
|
||
|
|
o last but not least relicensed from LGPL (>=2.1) to GPL (>= 2)
|
||
|
|
|
||
|
|
My blog (http://dirk.eddelbuettel.com/blog/) has two recent posts with a bit
|
||
|
|
more details (and colour highlighting of the code below) but let's just look
|
||
|
|
at one example of using GNU GSL functions for illustrative purposes (as you
|
||
|
|
wouldn't need this to access random-number generators as R has its own).
|
||
|
|
|
||
|
|
Consider this R code snippet:
|
||
|
|
|
||
|
|
## use Rcpp to pass down a parameter for the seed, and a vector size
|
||
|
|
gslrng <- '
|
||
|
|
int seed = RcppSexp(s).asInt();
|
||
|
|
int len = RcppSexp(n).asInt();
|
||
|
|
|
||
|
|
gsl_rng *r;
|
||
|
|
gsl_rng_env_setup();
|
||
|
|
std::vector<double> v(len);
|
||
|
|
|
||
|
|
r = gsl_rng_alloc (gsl_rng_default);
|
||
|
|
|
||
|
|
gsl_rng_set (r, (unsigned long) seed);
|
||
|
|
for (int i=0; i<len; i++) {
|
||
|
|
v[i] = gsl_rng_get (r);
|
||
|
|
}
|
||
|
|
gsl_rng_free(r);
|
||
|
|
|
||
|
|
return RcppSexp(v).asSexp();
|
||
|
|
'
|
||
|
|
|
||
|
|
## turn into a function that R can call
|
||
|
|
## compileargs redundant on Debian/Ubuntu as gsl headers are found anyway
|
||
|
|
funx <- cfunction(signature(s="numeric", n="numeric"),
|
||
|
|
gslrng,
|
||
|
|
includes="#include <gsl/gsl_rng.h>",
|
||
|
|
Rcpp=TRUE,
|
||
|
|
cppargs="-I/usr/include",
|
||
|
|
libargs="-lgsl -lgslcblas")
|
||
|
|
print(funx(0, 5))
|
||
|
|
|
||
|
|
|
||
|
|
Brief notes on this:
|
||
|
|
|
||
|
|
1) The character variable gslrng contains valid C++ code.
|
||
|
|
|
||
|
|
2) The call to cfunction converts this C++ code into a function that
|
||
|
|
calls it -- and this function is compiled, linked and loaded
|
||
|
|
automagically simply using hints about GSL header files and libraries.
|
||
|
|
|
||
|
|
3) The interface is explit: cfunction() is told about a function signature
|
||
|
|
with numeric variables 's' and 'n'; this is what the C++ code convert
|
||
|
|
to (C++ local) variables 'seed' and 'len' using the RcppSexp class.
|
||
|
|
|
||
|
|
4) We return a single SEXP to R which is based on the std::vector<double>
|
||
|
|
v which is also converted on the fly.
|
||
|
|
|
||
|
|
5) The resulting vector is returned and simply printed at the R level.
|
||
|
|
|
||
|
|
|
||
|
|
More examples are in the source tarball and in the R-Forge SVN archive.
|
||
|
|
|