Currently, we are storing things mostly in jsons. In workflow_meta.json the entire information about a step is stored. This includes parameter names, types, value ranges and information for dynamically filling forms with options.
Also, the logic for dynamically filling the forms is in run.py and run_helper.py.
Named outputs are also currently stored as python dictionaries, mapping keywords to dataframes or other datatypes to be used in other steps.
This is bad, and tedious to change.
The vision for this issue is to classify everything, meaning a hierarchical class structure for step, parameter, named_output and such.
The acceptance criteria have been talked about with Luca, but are hard to list because things are subject to change because this is such a big refactor. Please await the PR
Issue description
Oh boy, this is a big one...
Currently, we are storing things mostly in jsons. In
workflow_meta.jsonthe entire information about a step is stored. This includes parameter names, types, value ranges and information for dynamically filling forms with options. Also, the logic for dynamically filling the forms is inrun.pyandrun_helper.py. Named outputs are also currently stored as python dictionaries, mapping keywords to dataframes or other datatypes to be used in other steps.This is bad, and tedious to change. The vision for this issue is to classify everything, meaning a hierarchical class structure for step, parameter, named_output and such.
The acceptance criteria have been talked about with Luca, but are hard to list because things are subject to change because this is such a big refactor. Please await the PR