My complaint is really more that for this sort of thing I feel examples should be front and center; Jupyter "sells" a better experience, so show me how easy and ergonomic it is to use!
Hope that helps. Right now it looks like the sql parts are kind of the equivalent of running a subshell and displaying the results, with no communication to the "main" code. Is this accurate? Is there a way forward in the future to e.g. pull the data down with sql and then do some further munging in another language?
Hum, I see. I thought you haven't found the binder, sorry. It's a good point! I'll make sure to pay attention on that on the next posts :)
>"main" code. Is this accurate? Is there a way forward in the future to e.g. pull the data down with sql and then do some further munging in another language?
I'm not sure if I follow what you mean with the main code. you mean having access to each individual value of a query result, for example, to be able to manipulate it? if that's what you mean, that's what we receive in the C++ backend code that runs on this kernel. so as long as you can integrate new libraries to this code you'd be able to manipulate the results of your query.