I have taught a short-course and a seminar on an introduction to R for folks at the Natural Resources Research Institute and for the Quantitative Methods in Social Sciences here at UCSB, respectively. I espouse the free and open source nature of R as a virtue as well as access to hundreds (thousands?) of contributed packages available from CRAN (and of course packages not available on CRAN). Since contributed packages are not error checked other than the checks that CRAN makes, I always get asked the question about `how can you trust a package?' My answer to this question has been two-folds: 1) check it versus something that you know works and 2) if it has been around for a while then chances are that errors have been found and corrected. Any thoughts about other ways to respond? I thought about this recently when reading Christian Robert's blog pointing out a post about errors in the package `schoolmath'. This was actually reported on a blog post by Neil Gunther. The problem involves finding prime numbers, so be wary!

blog comments powered by Disqus


25 June 2010