DevInstruction.md

Steps to run the demo-lale.jl which is found in the demo subdirectory

1. Download Julia (https://julialang.org/downloads/)

• highly recommend to get the julia-v1.6.0(March 24, 2021)
• it will be released soon and has great features and faster compiler
• on Linux (including WSL = Windows Subsystem for Linux):

  > wget https://julialang-s3.julialang.org/bin/linux/x64/1.6/julia-1.6.0-rc3-linux-x86_64.tar.gz
  > tar xvzf julia-1.6.0-rc3-linux-x86_64.tar.gz
  > sudo cp -r julia-1.6.0-rc3 /opt/.
  > sudo ln -s /opt/julia-1.6.0-rc3/bin/julia /usr/local/bin/julia
  

• on Mac, after installing the image into the Applications folder:

  > sudo ln -s /Applications/Julia-1.6.app/Contents/Resources/julia/bin/julia /usr/local/bin/julia
  

2. At Julia prompt press ; to trigger Julia shell:

julia> ;
shell> cd /path/to/Lale.jl

Note: Press backspace to return to Julia prompt:

At Julia prompt:

julia> using Pkg
julia> Pkg.activate(".")   # activate the current environment
julia> Pkg.instantiate()   # install all the deps needed to run the package
julia> Pkg.update()        # get the latest versions of deps  
julia> Pkg.status()        # show all deps versions and status

You can also use the built-in package management shell to do similar steps above:

julia> ]                   # trigger pkg shell
pkg> activate .            # activate the current environment
Lale> instantiate          # install all the deps needed to run the package
Lale> update               # get the latest versions of deps  
Lale> status               # show all deps versions and status

Note: Press backspace to return to Julia prompt.

Remark: This will install the necessary deps by reading the Project.toml. Pkg.status() should list Conda, Pandas, PyCall which are packages needed for the demos. If they are not listed, you can install them manually. Always activate the current path containing the package you are working on which is equivalent to virtual environment in python. Any external package you install is locally resolved in the current package environment you are working and will not clash with other packages you are working on.

You can install a package such as PyCall by:

julia> using Pkg
julia> Pkg.add("PyCall")

or use IJulia Notebook

julia> using Pkg
julia> Pkg.add("IJulia")
julia> using IJulia
julia> notebook(dir=pwd())

You can also install kernels to run Julia with multithreading:

julia> using IJulia
julia> installkernel("Julia (4 threads)", env=Dict("JULIA_NUM_THREADS"=>"4"))

More info can be found in IJulia documentation.

You can also use Pluto Notebook:

julia> using Pkg
julia> Pkg.add("Pluto")
julia> using Pluto
julia> Pluto.run()

Remark: Julia REPL support tab completion. Pressing ] will switch to package management shell which allows you to install, query, manage, etc packages. Backspace returns to Julia shell.

3. Point PyCall to your python binary location:

julia> ENV["PYTHON"] = "/pathToPython/"
julia> Pkg.build("PyCall")

Remark: This will make sure it will use the correct python. You can also set ENV["PYTHON"]="" and PyCall will install python using Conda.jl. The location of python and pip3 is in $HOME/.julia/conda/3/bin/. You can add more python packages by using pip3 in the said directory.

4. Install python lale package

> cd /path/to/python/binary
> ./pip3 install lale

or

/path/to/python/bin/pip3 install lale

Remark: This will make sure lale is available during PyCall.pyimport("lale") in Julia

5. You can now try the julia lale demos in demo/ subdirectory.

6. You can load the Lale.jl package and try the demo-lale-package.jl in demo/ subdirectory

7. You can trigger unit-testing with:

julia> using Pkg 
julia> Pkg.activate(".") # activate current directory with Lale package
julia> Pkg.test()

or

julia> ]                 # trigger julia package shell
pkg> activate .          # activate lale package environment
Lale> test               # run Lale unit-test

Note: Press backspace to return to Julia prompt. Note: Instead of ], you can also do Pkg.action("arg")

8. To run the tests or samples from the CLI (without REPL), do this:

julia --project -e 'using Pkg; Pkg.activate("."); Pkg.test()'
julia --project demo/demo-lale.jl

9. Run Julia with the following argument to treat the current directory as a project:

> pwd
> path/to/Lale
> julia --project
julia> ]
Lale>      # indicates that Lale is the active project