default R personal library location is null

Question

I updated my Ubuntu 16.04 machine to R 3.4.1. When installing the first non-required package (eg, lubridate), I received the message:

would you like

Answer 1

On your computer, your current library is named after your R version.

For instance, my current lib is :

.libPaths()
[1] "/home/colin/R/x86_64-pc-linux-gnu-library/3.3"

as my current R version is 3.3.

So yes, every time you'll update R, you'll face this lib problem. Note that you can set the path to your old library with .libPaths(new = "path/to/your/lib") or copy and paste your old library in the new one.

Colin

Answer 2

You are close. The problem rests in the "commenting out" of those lines, made without checking compatibility. The same problem happens if you try to install a library manually within the REPL, eg using:

install.packages("survival")

With the difference that you get an "NA" error instead of "null".

Solution for future R upgrades, if you want minimal hassle:

• Restore the /etc/R/Renviron to the package default, so that it won't ask for your input (or just be overwritten) next time R is upgraded

• Add a Renviron in your home directory, eg $HOME/.Renviron, with the following content:

R_LIBS_USER="${HOME}/R/${R_PLATFORM}-library/3.4.1/"

Personally, each time R upgrades I reinstall all libraries with the new version. So I will modify that 3.4.1 with 3.4.2 or whatever new version I have, and then reinstall the libraries.

If you don't want to reinstall your libraries, you can try to remove the version subdirectory altogether, eg:

R_LIBS_USER="${HOME}/R/${R_PLATFORM}-library/

so that your old libraries will be immediately seen by R.

NB: I couldn't find a way to put the R version inside the Renviron, sadly, but this could be achieved using an .Rprofile instead (since that can contain R code).

Answer 3

I found that for me the most suitable solution was to edit /etc/R/Renviron.site:

# nano /etc/R/Renviron

And uncomment the line where it sets R_LIBS_USER environment variable.

Et voilà!

I can again use library(), require(), and install.packages() within R.