Add R Kernel to SystemLink Jupyter

[CFEARING01]

I would like to know if it is possible to add the R kernel to SystemLink Jupyter. Looking at the documentation for JupyterLab, this looks rather easy: https://jupyterlab.readthedocs.io/en/latest/user/documents_kernels.html

Is this possible, and has anyone used R in SystemLink Jupyter?

I have developed some R Scripts in R Studio and am wondering if I might be able to migrate them into Jupyter on my SystemLink Server to potentially automate the process and allow the data to be easily viewed.

Thank you

[JoshuaP]

Give this a try.

• R uses TEMP, which is not currently set on the environment spawned by Jupyter, to set TEMP change line 31 of C:\Program Files\National Instruments\Shared\Skyline\Python\3.7\Lib\site-packages\winlocalprocessspawner\winlocalprocessspawner.py from:
• win_env_keep = ['SYSTEMROOT', 'APPDATA', 'WINDIR', 'USERPROFILE']         to
• win_env_keep = ['SYSTEMROOT', 'APPDATA', 'WINDIR', 'USERPROFILE', 'TEMP']
• 2) Install R, if not already installed: https://www.r-project.org/
• 3) From an R console running as Admin (it is crucial to run as Admin, as we need the IRKernel installed for all users)
  
     2.1) Install IRKernel:
          install.packages('IRkernel') 
     2.2) Make the kernel available to SystemLink Jupyter installation (adjust the path under prefix as needed)
          IRkernel::installspec(user = FALSE, prefix = '"C:/Program\ Files/National\ Instruments/Shared/Skyline/Python/3.7"')

[CFEARING01]

Thanks Josh, I was able to get to step 3.1, but could not complete 3.2. Although running the R console did not install IRkernel for all users. I had to specify the lib:

install.packages("IRkernel", lib="C:\Program Files\R\R-3.6.1\library")

When I run IRkernel::spec I get the following error:

jupyter-client has to be installed but "jupyter kernelspec --version" exited with code 127. In addition: Warning message: In system2("jupyter", c("kernelspec", "--version"), FALSE, FALSE) : '"jupyter"' not found

[Alejandro_del_Castillo]

Hi, 

I believe the problem might be that jupyter-kernelspec.exe is not being found. As a workaround, could you try to use setwd from the R console to set the current directory to "c:\Program Files\National Instruments\Shared\Skyline\Python\3.7\Scripts" before running step 3.2?

[CFEARING01]

That did the trick! Let me document the steps I took in case anyone else comes along

1. R uses TEMP, which is not currently set on the environment spawned by Jupyter, to set TEMP change line 31 of C:\Program Files\National Instruments\Shared\Skyline\Python\3.7\Lib\site-packages\winlocalprocessspawner\winlocalprocessspawner.py from:
   win_env_keep = ['SYSTEMROOT', 'APPDATA', 'WINDIR', 'USERPROFILE'] to
   win_env_keep = ['SYSTEMROOT', 'APPDATA', 'WINDIR', 'USERPROFILE', 'TEMP']
2. Install R, if not already installed: https://www.r-project.org/ (I used R-3.6.1)
3. Navigate to C:\Program Files\R\R-3.6.1\bin
4. Right click R.exe and run as administrator - Opens an R Console
5. Install IRkernel (define lib to install for all users)

6. install.packages("IRkernel", lib="C:/Program Files/R/R-3.6.1/library")

7. You will be prompted to select a CRAN mirror. I chose USA (CA)
8. Load IRkernel library

9. library(IRkernel)

10. Set working directory

11. setwd('c:/program files/national instruments/shared/skyline/python/3.7/scripts')

12. Install Kernel

13. IRkernel::installspec(user=FALSE, prefix='"c:/program files/national instruments/shared/skyline/python/3.7"')

14. Launch SystemLink Jupyter Notebook. You should now see the R Kernel appear in the launcher