Depends on what you mean by SQL app. Can you please elaborate more?

I've lately settled for a semi-functional approach with drf - each action is a pipeline, and you supply functions to it composing final behaviour

Writing like this is such a joy, you just implement several functions and plug it in

Also allows to easily opt out of the batteries when needed

This is also my main inspiration for the jikan library tbh - most ways of dealing with dates and time in Python are kinda bad, so half of the functionality of jikan is just copy and paste from my production code (stripped of any domain-specific parts of course)

Oh my god

May I suggest using some smaller models for some steps of this pipeline? Sentiment analysis with ChatGPT looks at least excessive

Cmon it's not thaaaaat bad

Morning nanami!

Making a schedule app for a private school + time-range manipulation library

To be honest, django type stubs are a mess (mostly due to QuerySet API inheritance structure being a mess) - most times it's either wrong or oblivious and it messes up LSP pretty badly

For example - I have a M2M relationship, and type stubs are suggesting parameters to `add` that simply don't exist