Can't register the kernel in the current R installation.

[darkRaspberry]

• I followed the official installation instructions
• I reproduced the error with the newest versions of IRkernel, IRdisplay, and repr
• I included a minimal reproducible example
• This the right repository: I think the issue is with IRkernel, not
  IRkernel/repr or a third party repository

What's issue?

I can't register the kernel in step 2.

What I have done.

1. I started with installing juypter as usual from official docs and run pip install jupyterlab HERE, I installed jupyter-lab not classic notebook in my system (Ubuntu 23.04) and after some issue with python 3.11 I installed it anyway.

2. I installed R by compiling it from source. And installed (R version 4.3.0 (2023-04-21) -- "Already Tomorrow")

3. I then launched R and installed the package install.packages('IRkernel')

4. I tried registering and it just didn't completed. And through this error.

> IRkernel::installspec()
Error in IRkernel::installspec() : 
  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) :
  error in running command

This seems juypter-lab is not installed but I tried running juypter and it works normally.
I reinstalled juypter-lab but nothing happened.

➜  ~ pipx install jupyterlab
Installing to existing venv 'jupyterlab'
  installed package jupyterlab 3.6.3, installed using Python 3.11.2
  These apps are now globally available
    - jlpm
    - jupyter-lab
    - jupyter-labextension
    - jupyter-labhub
done! ✨ 🌟 ✨

[darkRaspberry]

Further running jupyter --version shows no results and only jupyter-lab --version show anything. Installing classic notebook also doesn't help.

➜  ~ pipx install notebook
  installed package notebook 6.5.4, installed using Python 3.11.2
  These apps are now globally available
    - jupyter-bundlerextension
    - jupyter-nbextension
    - jupyter-notebook
    - jupyter-serverextension
done! ✨ 🌟 ✨

still I see no version with jupyter --version
But running jupyter-notebook --version shows it correctly installed.

[flying-sheep]

The error message says:

> IRkernel::installspec()
Error in IRkernel::installspec() : 
  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) :
  error in running command

Did you read the warning here? https://irkernel.github.io/installation/

On OS X, be sure to execute this in R started from the Terminal, not the R App!
(This is because the R app doesn’t honor $PATH changes in ~/.bash_profile)

You probably executed that command in an environment that doesn’t have jupyter-kernelspec in its PATH.
Make sure things work before running IRkernel::install():

$ jupter kernelspec -h
Usage: jupyter kernelspec […]
[…]
$ R
> IRkernel::install()

[sje30]

I wonder if the problem here is that 'jupyter-lab' binary is installed by jupyterlab, whereas the R package assumes the binary is called 'jupyter'. (this at least the problem I'm experiencing).

[flying-sheep]

Almost. The package jupyterlab provides a shell command called jupyter-lab. You shouldn’t use this shell command directly, you should use the jupyter shell command provided by the jupyter-core package. jupyter lab will call jupyter-lab internally, and that’s the preferred way to call that.

jupyter-core is a dependency of jupyterlab, therefore in a properly configured environment, you will have both shell commands available.

The only exception I can think of is pipx, since by default, pipx doesn’t export shell commands of dependencies. So if you use it to install jupyterlab without specifying --include-deps, it will only make jupyter-lab but not jupyter available in the shell.

[sje30]

Thank you for the explanation.. I will need to investigate what I installed. I am on Mac and so used brew (+ also use pyenv to manage python versions).

[sje30]

Just to follow up... I think my problem was that pyenv didn't install Jupyter.

on my Mac, if I do:

brew install jupyter

then all is okay.

thanks again for clarifying.