A story that solved the problem that the Java (jdk) version was too high to use the h2o library in R and R Studio.

Motivation

If you think you wrote it after a long time, it's about PATH again. For Mac. I'm sorry for Win people. An angry note that solved the problem that the jdk version was caught and the problem did not proceed when the problem of using the deep learning library called h2o in the R environment came out at school.

Solution

Lower the version of jdk used under R environment. The following is a specific method.

What's in Java now?

Launch a terminal instead of the R console and

/usr/libexec/java_home -V

Let's type in and find out. All included Java versions will come out. It's like this.

Do you have a jdk that can use the h2o library?

The h2o library works with the jdk. At this time, it seems that the lever does not work unless the version of jdk is between 7 and 12. The latest version of jdk is 13! !! !! !! !! !! !! !!

If you don't have the target version of the jdk, get one

If you didn't find the target version when you checked the Java version, download and install it. You can get it from the official website (LTS 11 in the middle instead of 13 that stands out at the top Let's DL).

If you get it, just in case, at the terminal again

/usr/libexec/java_home -V

Please see if you entered properly.

Let's lower the version of jdk

Please be careful here, but the path of jdk used when playing with Java programs in the terminal and the path of jdk referenced by the R environment are different. If you use the following method, the version of jdk used normally will remain 13, and the version of jdk used under the R environment will be 11. You did it.

Search for Renviron files in Finder

Renviron files are like .bash_program in R environment. If you have R (and RStudio) installed, you should have this file. Open it in a suitable editor and even at the bottom

JAVA_HOME=/Library/Java/JavaVirtualMachines/openjdk-11.0.2.jdk/Contents/Home

Let's write.

Done! !!

that's all. The jdk referenced by R is now downgraded. On the R console

python

library(“h2o”)
localH2O = h2o.init(ip = "localhost", port = 54321, startH2O = TRUE, nthreads = -1)

But let's hit it and use h2o. It is said that the version of h2o itself is old, but it can be used.

Thank you for your hard work.

Postscript

The currently distributed LTS11 seems to be 11.0.6 instead of 11.0.2. Please check the version you got and change the PATH each time.