Hi folks - Sylvain Gugger and I created nbdev. Let me know if you have any questions about it.
Since nbdev is written in nbdev itself, you can also use its repo as an example project to see how it works. Another good example project written with nbdev is fastprogress: https://github.com/fastai/fastprogress
Looks great ! I don't know If you use Apache Airflow in ML or ETL pipelines but I wonder if one could use nbdev to generate Airflow Dags/Operator. There's already some work done to execute Notebooks[0] but would be cool to convert them to Python code!
So is the ultimate goal here to have something highly annotated? I.e. instead of having programmers comment their code, you have them code their comments?
I like the principle, but what you say are the limitations so far? What can it not do?
Since nbdev is written in nbdev itself, you can also use its repo as an example project to see how it works. Another good example project written with nbdev is fastprogress: https://github.com/fastai/fastprogress