Daily I'm working with a lot of SQL, mostly in a self-hosted Redash. But, I caught myself on commenting some part of a query and then writing another query below the commented-out code. I thought it would be awesome to work with data in a Jupyter-like fashion - then I could tell the whole story, why I went this way with research, why such a query. So I added SQL support to my Jupyter instance, a started working with data in Jupyter.

Now I can tell a better story and I have the full history of "why I did this". But it feels somehow not cool, also keeping the credentials to the database inside the notebook - not cool. _(I know I can keep them in environment, but that's not the point)_

Is anyone querying SQL from Jupyter? Is there a better way to query data, and query the results with another SQL? Then use Python to do some stuff on data. Then plot the results. Keeping all the mid-steps at the same time.