-
Notifications
You must be signed in to change notification settings - Fork 10
Open
Description
I've been playing around with using LLM's to generate/convert docstrings and found two docstring formats for Python that are widely used:
- PEP 287 (https://peps.python.org/pep-0287/) reST and Sphinx compatible but slightly boring
- Numpydoc (https://numpydoc.readthedocs.io/en/latest/format.html) reST and Sphinx compatible (with plugin) and slightly more sophisticated when it comes to dealing with code examples
I'm considering both to see if there are any quality differences in the AI output but perhaps we should, in general, select one for PySCeS?
Update. It looks like Numpydoc format gives more consistently formatted output so that would be my suggestion to use for new docstrings.
Metadata
Metadata
Assignees
Labels
No labels