R/Java interface allowing the use of Java from R as well as embedding R into Java (via JRI)
Please visit the main rJava project page on RForge.net for details on the project. For some FAQs and troubleshooting see below - read before reporting bugs!
Recommended installation of the CRAN version is via
install.packages("rJava")
in R. If you have all tools (and knowledge) necessary to compile R packages from sources, you can install the latest development version with
install.packages("rJava", repos="https://rforge.net")
The RForge.net repository is updated automatically on each
commit. On macOS/Windows you may need to add type='source'.
IMPORTANT: You must have Java installed and it must be of the same architecture as the R you are using. See below for some troubleshooting help.
When checking out the sources, you must use
git clone --recursive https://github.com/s-u/rJava.git
since rJava includes REngine as a submodule. If you want to create a
package from the source checkout, you must use sh mkdist to do so
since the checkout is not the actual R package but a source to
generate one (which involves compilation of Java code).
Please use rJava GitHub issues page to report bugs, but read the following documentation and search previous issues before you do so.
Rule #1: do not set JAVA_HOME unless you are an expert. rJava attempts to find the correct settings automatically on most platforms, so setting JAVA_HOME incorrecty will just break things.
Please make sure you install Java that matches your R architecture. R from CRAN is installed by default both in 32-bit and 64-bit versions so if in doubt, install both 32-bit and 64-bit Java. The most common mistake is to use 64-bit R but only have 32-bit Java installed.
rJava determines the Java location from the registry, so make sure you use the official Oracle installer so that your Java installation can be found.
On modern macOS versions Apple no longer supplies Java, so it must be downloaded from 3rd parties. Probably the most commonly used distributions on macOS are adoptium.net and Azul Zulu. Please note that if you are using arm64 R on Apple silicon (M1+) based Macs you will need at least R-4.1.2 or else you will get trap R errors when loading Java (see #267 for details).
When installing from a zip or tar ball, put your Java installation in /Library/Java/JavaVirtualMachines. For example, if installing Zulu, unpack/move it such that it results in /Library/Java/JavaVirtualMachines/zulu-11.jdk.
Most recent rJava version will try to automatically detect the Java location and load it dynamically. You can also check the version selected by your settings via /usr/libexec/java_home in the Terminal.
If you have multiple versions and want to pick one without changing the macOS Java settings, you can set JAVA_HOME but it must point to the Home directory inside the JDK, so, for example, for that above Zulu JDK that would be JAVA_HOME=/Library/Java/JavaVirtualMachines/zulu-11.jdk/Contents/Home. Again, don't do this unless you want to change the default behavior.
There is no standard location of JDK on Linux, so you must configure R with Java support before you can use rJava. It is usually done by running R CMD javareconf which detects all necessary settings and modifies the Java configuration in $R_HOME/etc/javaconf. Note that you must have sufficient privileges to update that file in order to configure R.
Also note that sudo may change environment variables, so if you need to run with elevated privileges, try sudo -i first then check if you still have access to the Java you want to use and then run R CMD javareconf. Alternatively you can temporarily change permissions on $R_HOME/etc to allow you to update it.
The way Java R configuration on Linux works is for the R start script to modify LD_LIBRARY_PATH to make sure the JVM libraries can be loaded (it does so according to the javaconf settings). Therefore if you use a process embbedding R you need to run it via R CMD <program> such that those setting are honored, otherwise you're on your own.
If you are installing rJava from sources, make sure you have the full JDK installed and all the necessary libraries needed to compile packages. For example, on Debian/Ubuntu that would require at least r-base-dev. If you run into issues, please check config.log which gives a clue as to what went wrong - usually some missing R dependency such as pcre2. The config.log file will be in the directory you used to build rJava in which is claned by R by default, so to keep it you can use e.g.:
curl -LO https://rforge.net/rJava/snapshot/rJava_1.0-6.tar.gz
tar fxz rJava_1.0-6.tar.gz
R CMD INSTALL rJava
## on failure check rJava/config.log