Cool project. There is lots of interesting work going on around bridging the gap between notebook analysis and production. I think the SQL generator is pretty useful there.

How syntactically different is `rql.dataset` object from a `pandas.dataframe`? Would it also be possible to have your .sql() function translate existing pandas operations to sql?

thanks! we initially wanted syntax to be identical but later realized it doesn't make sense because dataframes rely on indexes for a lot of operations. i do think we can eventually support direct translation of some pandas functions though.

the cool part about rql.dataset is it's only a metadata pointer to the dw... so it doesn't pull any physical data into RAM

Neat project! This looks like it'll come in handy quite a bit in my day to day.

I'm still looking through the readme, but I have a question. When I build models, sometimes I pair the modeling with Standardization or log transforms - is RasgoQL the right place to be doing those types of transformations?