Abstract:
|
Many statistical software packages come as a base system providing core functionality (a language plus a run-time environment with graphics, and ``basic'' statistical facilities), and extensions provided in particular by the user community. For R, the open source implementation of the award-winning S language, the Comprehensive R Archive Network (CRAN, http://CRAN.R-project.org/) acts as a central repository for material related to R, and currently contains more than 140 contributed add-on packages. To ensure feasibility and success of such a two-tier model for development and deployment, it is necessary that add-ons can easily be created, installed and used. To this end, R comes with two tools for building and checking add-on packages. The latter performs a variety of tests, including availability of required meta-information, out-of-the-box installability, absence of common coding problems, completeness and consistency of documentation, and general (via executable examples) and specific run-time tests. We present these tools in some detail and discuss the range of applicability of the underlying ideas and the effect on maintaining quality standards within the R system.
|