Julia Kernel about ml-workspace HOT 6 CLOSED

Thanks for your suggestion. We currently do not have any plans to add a dedicated Julia flavor image. However, we have the concept of "on-click" installer scripts. A julia/julia kernel installer script would be a fantastic addition for the workspace. Currently, we do not have the resources to add this on ourselves, but if you want to try to contribute such a script you can take a look at this installer script for the r-interpreter/kernel as inspiration. Keep in mind that within the workspace you have full freedom to install anything that runs on ubuntu, e.g. you can install the julia interpreter via: apt-get install julia.

from ml-workspace.

Okay thanks! I've tried to put together a script based on the r-runtime installation script. I'm open to feedback if it's something you'd consider adding to the tools but can also just use it myself if it's not worth the time from your point of view.

#!/bin/sh

# Stops script execution if a command has an error
set -e

# Loop through arguments and process them
for arg in "$@"; do
    case $arg in
        -i|--install) INSTALL_ONLY=1 ; shift ;;
        *) break ;;
    esac
done

if ! hash julia 2>/dev/null; then
    echo "Installing Julia runtime. Please wait..."
    # See https://github.com/jupyter/docker-stacks/blob/master/datascience-notebook/Dockerfile
    
    # environment variables
    export JULIA_DEPOT_PATH=/opt/julia
    export JULIA_PKGDIR=/opt/julia
    export JULIA_VERSION=1.2.0
    
    mkdir /opt/julia-"$JULIA_VERSION" && \
    cd /tmp && \
    wget -q https://julialang-s3.julialang.org/bin/linux/x64/`echo "$JULIA_VERSION" | cut -d. -f 1,2`/julia-"$JULIA_VERSION"-linux-x86_64.tar.gz && \
    echo "926ced5dec5d726ed0d2919e849ff084a320882fb67ab048385849f9483afc47 *julia-$JULIA_VERSION-linux-x86_64.tar.gz" | sha256sum -c - && \
    tar xzf julia-"$JULIA_VERSION"-linux-x86_64.tar.gz -C /opt/julia-"$JULIA_VERSION" --strip-components=1 && \
    rm /tmp/julia-"$JULIA_VERSION"-linux-x86_64.tar.gz
    
    # symlink /usr/local/bin/julia to /opt/julia-$VERSION
    ln -fs /opt/julia-*/bin/julia /usr/local/bin/julia
    
    # Show julia where conda libraries are
    mkdir /etc/julia && \
    echo "push!(Libd1.DL_LOAD_PATH, \"$CONDA_DIR/lib\")" >> /etc/julia/juliarc.jl && \
    # create JULIA_PKGDIR \
    mkdir $JULIA_PKGDIR
    
    # Install IJulia and then move the kernelspec out
    # to the system share location.
    julia -e 'import Pkg; Pkg.update()' && \
    julia -e 'import Pkg; Pkg.add("HDF5")' && \
    julia -e "using Pkg; pkg\"add IJulia\"; pkg\"precompile\""

else
    echo "Julia runtime is already installed"
fi

from ml-workspace.

Hey @sterlinm, Thanks for your contribution 👍 I would definitely prefer to add this script to our tools collection. I just tested the script and installation works fine! Using Julia from command line also works for the limited examples I have tested. However, the Julia Jupyter Kernel (that was added during installation) is currently unable to start:

Stacktrace:
 [1] require(::Module, ::Symbol) at ./loading.jl:876
 [2] include at ./boot.jl:328 [inlined]
 [3] include_relative(::Module, ::String) at ./loading.jl:1094
 [4] include(::Module, ::String) at ./Base.jl:31
 [5] exec_options(::Base.JLOptions) at ./client.jl:295
 [6] _start() at ./client.jl:464
in expression starting at /opt/julia/packages/IJulia/fRegO/src/kernel.jl:1
[I 13:17:48.351 NotebookApp] KernelRestarter: restarting kernel (1/5), new random ports
ERROR: LoadError: ArgumentError: Package IJulia not found in current path:
- Run `import Pkg; Pkg.add("IJulia")` to install the IJulia package.

Also, you can remove the && \ line endings from the script, they are only required if used within a Dockerfile.

If you like, you can create a julia-interpreter.sh script in the tools folder and contribute it via pull request. As an alternative, I can also add the script myself for the next version.

from ml-workspace.

Hey @LukasMasuch sorry for the delayed response. I'm happy to do a pull request, though I want to test the Jupyter kernel issue first. I'm able to create a jupyter notebook with the Julia kernel and the kernel starts fine. I'll try starting a new instance from scratch with the script to make sure it wasn't some manual tweak that I forgot about.

from ml-workspace.

@sterlinm were you able to fix the issue?

from ml-workspace.

This issue is stale because it has been open 90 days with no activity. Remove stale label or comment or this will be closed in 14 days

from ml-workspace.