Moving or copying the library when upgrading R

Because I need to be sure that my packages work with current versions of R, I nowadays cannot skip any R update. I use many different packages, so re-installing all the packages in the library is something I would rather avoid.
Today I found that an easy way to automate this task is to use package installr, more specifically, a function with a very long, but descriptive name copy.packages.between.libraries().

copy.packages.between.libraries(ask = TRUE, keep.old = TRUE)

Will ask you to choose source and destination libraries from a list of available ones and copy without deleting them from the old library all packages. Using keep.old = FALSE, the default, will move the packages, and ask = FALSE the default will automatically select source and destination. This automatic guessing did not select the direction I expected for the case of a patched versus released version of R.
Anyway this is handy when one has, as the function itself reported some minutes ago 144 packages to move!
If you do not use patched versions of R this should be enough.


I hope this bit of advice will make less painful my decision of no-longer supporting R 3.1.X and earlier!


RStudio (new preview)

The people developing RStudio have done it again!

I have been using for a couple of days RStudio from the 0.99 version previews, and it is working extremely well. The new functionality improves very much the easy of use: now there is auto-complete and automatic “bubble” help, both are a hugely helpful when writing scripts and package code.

You may still want to wait, as new versions are being released almost every other day. Anyway, I wanted to share the good news.

data.table 1.9.4 available from CRAN

A new version of the data.table package was released by its maintainers to CRAN. This version should work with the current version of the photobiology packages. Users of my packages should update data.table. At the moment a binary package for OS X seems not to be available, until it becomes available, you will need to install from source on Mac computers, which requires instalation of Xcode and accepting its license. For Windows, binary is available from CRAN.

ggplot2 version 1.0

ggplot2 version 1.0 has been released last week. Last March the author of this package declared a “feature freeze”. This means that no new functionality will be added in the future, although the package will continue to be maintained and kept working with future versions of R. No changes to our suite of packages or its documentation were triggered by this update.

In the future extensions to ggplot2 will be in separate packages. Two good examples are ggmap and ggtern. ggmap can be used to plot data (using regular ggplot syntax) on top of a map. ggtern adds functions for plotting ternary plots.