As announced this morning on mailing list, R 3.1.0 (codenamed "Spring Dance") has been released. The source code is available now; as of this writing binary versions haven't yet appeared on CRAN or propagated to the mirrors, but I expect they'll be available in a day or two.
You can check out the full list of changes in the Release Notes, but the highlights include:
- Better big data support for some matrix summary operations and the Fast Fourier Transform.
- A new function anyNA for quickly detecting missing values in a vector.
- Improved timezone support and better support for dates in the far past and future.
- New support for package developers to include C++11 code.
- Improved accuracy for the far extremes of some statistical distributions, and minor bug fixes for a few functions.
- Reduced memory use (for example, fewer copies made of objects during some assignments). Intriguingly, R now supports a reference counting system which promises to reduce memory use even further. It's not used by default in 3.1.0, but it "may become the default in the future".
The R-statsitics blog has some tips on upgrading from a previous release of R. R 3.1.0 is the latest annual update to R. There are likely to be patch updates in the interim, but if the past is any guide, version 3.2.0 will be out in April 2015.
r-announce mailing list: R 3.1.0 is released