R Core member Professor Brian Ripley from Oxford University gave the first keynote presentation of useR! 2011 today, and gave some insights into what goes on behind the scenes to create two updates to R (plus several patches) every year. He began with some facts about the history of R (noting that if they'd known R would take off like it has, there would be better records of the early days):
- The first still-existing version of R dates from Jun 1995, and the distribution totals 465Kb.
- R 1.0.0 was released on February 29, 2000 (up to 2.8Mb)
- R 2.0.0 was released on October 4, 2004 (mostly because the name R 1.10.0 was unappealing because it would sort to the top of the list of R versions in some systems). At this point the distribution had grown to some 10Mb in size.
Prof. Ripley also showcased some of the major improvements from recent versions of R, including multi-language support (thanks to which R is widely used in China and Japan, for example), support for R as a scripting language (most of R's own build scripts are now written in R), and improved graphics rendering.
Looking to the future, Prof. Ripley noted that there are no plans to make any backwards-incompatible changes that would warrant a jump to a 3.x numbering scheme. R 2.14 is planned for October, and after that the Core Group will move to an annual (rather than bi-annual) release schedule beginning with R 2.15 in (provisionally) March October 2012. [Updated Nov 17 2011 -- my error.] He also gave a glimpse even details of R's development plans, with low-level support for multi-threaded computing, a standard parallel computing library, and support for a 64-bit native R engine possibly on the horizon.
The talk also included some rather poignant insight into the level of altruistic commitment provided by the active members of R-core to keep the R project running as smoothly as it does. For example, there are more than 110 contributions to CRAN each week, each of which requires manual review and often direct feedback on how to fix problems from CRAN maintainer Kurt Hornik. Also, many members of R-core spend a lot of volunteer time on the R-help and R-devel mailing lists interacting with R users: so many request for help and suggestions for changes to R take a lot of effort to respond to, even when asked respectfully -- and these contributions from R-core perhaps aren't always treated with respect.
So I'd like to join with the rest of the R community in giving thanks to Prof Ripley and the R core team for making R available to the community at large. Each of us has benefited greatly from their selfless contributions in taking statistical computing to the next generation and I, amongst many I'm sure, am extremely thankful to them for their generosity.
useR! 2011 Invited Talks (abstracts): The R Development Process, Brian Ripley