Building tools with NBDev - Making Python packages via Jupyter notebooks.

We're experimenting with live streams!

In this live-stream Vincent builds a plugin for SQLite-vec but he will do this in a special way. Instead of doing everything via Python files, he will do it all via Jupyter by using NBDev. It turns out that you can actually build packages this way, but you will have to rethink a few things. The goal of the sessions is to go over some pros and cons of the approach while also talking about Python packages in general.

The code for the VectoyLite repo can be found here:

We also host a podcast called Sample Space, which you can find on your favourite podcast player. All the links can be found here:

If you're keen to see more videos like this, you can follow us over at @probabl_ai.

[jeffrey]

i love nb dev and started to push to use it in my team. The only complaints were that diffs for NBs kinda suck and we are on azure devops and it security does not allow us to install the NB Viewer extension. I still think that these are minor concerns as you still have diffs from the generated py files for checking the code.
But the upside is just 10x more easier to understand code bases with ease of adding documentation and also so easy to just open the nb and play around with the functions. Also you get at least always a bare minimum of testing for free instead of sometimes zero testing (especially during mvp stage)