Local 'language' spreads worldwide

By RICHARD WOOD

An open source software project started 10 years ago in New Zealand is now in use by tens of thousands of academics globally, and a foundation has been established to ensure its continuing success.

The programming language, called R, is mostly used by academic statisticians and was founded by Ross Ihaka and Robert Gentleman at the University of Auckland's department of statistics.

Ihaka said the language began in 1992 as just trying out a few ideas.

"We decided to write something for our first-year course and made it look like a language called S, produced by Bell Labs."

That attracted people's attention and it grew. Hundreds are contributing and the group even has a member from Bell Labs.

"We've essentially got all the best researchers working on it now in the statistical computing area," said Ihaka.

He and Auckland University colleague Paul Murrell are part of a core group of 20 who can modify the code.

The software is licensed free under the open source GNU licence and development has been co-ordinated through the website www.r-project.org since 1998.

Exact numbers of users are not known because the software is freely down-loadable. It is available in Linux, Windows and Macintosh versions.

Competing software includes the original S language, its development Splus, and statistical packages such as SAS and SPSS. A Java-based project called OmegaHat is related in the sense that many contributors work on both projects.

R is primarily a command language-driven tool, but a number of projects are in train to add a graphical interface to it.

Ihaka said the next stage of development could be adding distributed computing support and the ability to plug in components.

Development is continuing on object-oriented extensions, the ability to compile for a 20 to 30 times performance improvement, and extensions for visualisation.

A non-profit foundation called the R Foundation for Statistical Computing has been formed and $40,000 raised so far to pay for development beyond the traditional open source model for both R and other projects.

Ihaka said his primary interest in R now was to build software that did useful graphics.