Implement adjustable base environments / package management for publishing process
Problem to solve
As a user I want to be able to adjust the libraries I require for my scripts easily. For example, I created this repository: https://mlreef.com/mlreef/cloob
The script iself requires several libraries and packages that are outside of pip. I attached a yaml file that uses conda to install any dependencies i specify in this file: https://mlreef.com/mlreef/cloob/-/blob/branch/master/path/environment.yml
Now, as a user I want to have it as easy as possible to port this script to mlreef, without having any limitation.
As a solution, we could provide conda (or mini-conda) in the base enviornment and load the environment.yml file during publishing process eg. with this additional line
conda env create -f environment.yml
(if the user selected this method / or do it automatically -> more review of the process is needed)
Proposal for Technical Solution
Permissions and Security
Documentation
Some good comments on conda dependency installing / parsing to requirements.txt: https://stackoverflow.com/questions/56700687/installing-dependencies-from-conda-environment-yml-without-conda