On 11/17/2010 01:57 PM, Dirk Eddelbuettel wrote: > > On 17 November 2010 at 10:44, Rainer M Krug wrote: > | On 11/17/2010 10:26 AM, Romain Francois wrote: > | > Ultimately, making a package is the best action. > | > | True - But I haven't looked at that option yet, as I was happy with > | sourcing the code, but I will for my next project. > > Are you sure you haven't? ;-)
Not for this project - but that package in which we are co-authors definitely is doing that. To clarify: Up to now, the whole simulation (of which the the code discussed here is s small part) was purely R, as it worked fine and was relatively easy to maintain. Now the simulation is growing, and the need for speed becomes more evident. So I am looking into the coding sections in C++ and to use inline or Rcpp. When the C++ code is working, and before the simulations can run on a hpc cluster (I don't suppose gcc is installed on the nodes), I have to compile it and the idea of a package, also in relation to version control, is really appealing. But that is the next step. Don't worry - I haven't forgotten earthmovdist and I will use the code there as a blueprint for wrapping the C++ code. > > You and I are co-authors of the earthmovdist package on R-Forge which is just > that: a small package with Rcpp around a single file you once for help with > in wrapping to C++. Feel free to take it as a stanza. > > https://r-forge.r-project.org/projects/earthmovdist/ > Cheers, Rainer > Dirk _______________________________________________ Rcpp-devel mailing list [email protected] https://lists.r-forge.r-project.org/cgi-bin/mailman/listinfo/rcpp-devel
