Closed turbomam closed 1 month ago
Normally I would say "add a requirements section to the README", but your README already has a god flow to it
You could make a one-line requirements.txt
and assume people know how to use that with pip install -r
Just to be clear, do you mean instructions for installing make_data.py so it can be run in any directory? If you have an exact requirements.txt line I should add I would appreciate that.
My original intent was just to communicate that users have to install the dpath
module.
The ability to run make_data.py
from any template directory without specifying the "full relative" path would be nice, but I wasn't asking for that.
Ah right! Ok, good, I'll add that requirements.txt into our repo too.
I've merged your patch. Not sure a version needs to be mentioned for dpath - the script uses the basic dpath stuff.
Might it be worth making a "Quick Start script/make_data.py" page in the wiki for new users?
Might it be worth making a "Quick Start script/make_data.py" page in the wiki for new users?
Yeah. I can help with that if I have sufficient permission.
The requirement for dpath
is semi-documented in the linkml-datastructure
's requirements.txt
now. We would just need to add the standard
python3 -m venv venv
source venv/bin/activate
pip install -r requirements.txt
Then the user would add their own template folder and LinkML source files and write a Makefile rule based on the provided samples.
Likewise,
I will close this now. Our code has changed a bit since make_data.py; now it is called tabular_to_schema.py and is run in a particular schema folder (see https://github.com/cidgoh/DataHarmonizer/wiki/DataHarmonizer-Templates). Soon a change will be implemented so that people can work on templates independently of a DataHarmonizer source code installation.
More info on this, or close it?