Duncan J. Murdoch (PhD in Statistics, Carleton University) is a professor in the Department of Statistical and Actuarial Sciences at Western University, London, Ontario. His interests are in applied statistics (especially orientation statistics) and statistical computing (especially the R project and perfect sampling). Dr. Murdoch is a member of the R core development group.

R workshop: Debugging and package writing

Presented by Duncan J. Murdoch (Western University)
Date : June 9, 2014, 11:00-12:00 and 13:30-15:30
Lieu : Krieble Lab, Room 3120, Trottier Building, McGill University

Sponsored by the CRM and the Montreal Biostatistics Seminar Series

Debugging in R (11:00-12:00)

In this tutorial, Dr. Murdoch will describe the support in R for debugging scripts and functions. He will start with the classical methods of printing values as code runs, and progress through methods triggered by errors or warnings, up to setting of breakpoints in code. Most of the tutorial will be platform-neutral but at the end, Dr. Murdoch will demonstrate how recent versions of RStudio take advantage of the breakpoint features to allow visual debugging.

Writing R packages (13:30-15:30)

R has become successful in part because of R "packages", modules of code and documentation, mainly written by R users. In this tutorial, Dr. Murdoch will describe why packages are a good way to develop R code even if you don't intend to distribute it to others. He will start with simple packages containing just R code, and will progress to more complicated packages including C (or other) compiled code. Most of the issues here are common to all platforms: how to set up the package, how to install, test and debug it, and how to package it for distribution to others. Short descriptions will also be given of how to set up Mac OS and Windows systems for package development.