Use nbdime for PySpark Notebooks

[gr7ffo]

would it be possible to use nbdime/ndbiff inside PySpark Notebooks? Often these type of notebooks have a specialized jupyter notebook format, e.g. Azure Synapse Analytics Notebooks.

[mmaitre314]

I was thinking potentially a hybrid between nbdime and the Jupyter Notebook Renderer, at least to enable preview of Synapse notebooks in Azure DevOps (ADO). Couple of challenges I hit so far in that scenario:

• ADO only allows client-side diff (i.e. HTML/CSS/JS and no Python). I tried compiling nbdime to WebAssembly but the dependencies are too complex and it failed.
• Synapse notebooks use the generic .json extension so it is not obvious for which files to enable notebook preview (e.g. enabling notebook preview on ARM template JSON files would be confusing...).

(that's assuming ADO would be a good place to address the feature-request above)