Julia dependencies needed for taxdata
After @andersonfrailey's helpful responses in #373 and #375, and additional changes related to julia that I describe below, make cps-files now appears to be working on my ubuntu system, although it looks like it could take a few hours.
I ran into some trouble so here is documentation, for others, of how I addressed it.
On my first effort at make cps-files after adding __init__.py to the cps folder and putting CPS data in the data folder, I ran into errors with Julia. I am using julia version 1.4.1. It turns out several required julia packages were missing which I learned with each new run of make cps-files. Also, there was a weird julia wrappers error. Here is a consolidated list of the steps I took to address this.
Starting in a ubuntu terminal, with the conda environment taxdata-dev activated, I did the following. I don't know much about julia so there might be better ways.
(NOTE: I show the terminal prompt as "...taxdata$" and the julia REPL prompt as "julia>".)
...taxdata$ julia # enter julia REPL
julia> using Pkg # fire up the package manager
julia> Pkg.instantiate() # I think this is supposed to install needed dependencies for taxdata-dev but it didn't get them all
# install additional needed packages
julia> Pkg.add("JuMP") # optimization modeling language; note the mixed upper and lower case - must be exactly as shown
julia> Pkg.add("Cbc") # interface to the COIN-OR Branch and Cut solver; same comment wrt mixed case
julia> Pkg.add("NPZ") # to read and write numpy array files
julia> exit() # get out of julia
...taxdata$
This was almost everything. However, I was getting this error: "LoadError: ArgumentError: Package Bzip2_jll does not have JLLWrappers in its dependencies" .
On a guess, after reading this, I used my system file manager to delete two files that julia had created in the taxdata folder: "Project.toml " and "Manifest.toml". There may also be ways to do this from within julia with Pkg.resolve() but I did not try that.
With these changes make cps-files seems to be working but it is too early to be 100% sure.