I just had a neat idea, recently we added support for auto parsing of sql exprs for

where

ex:

df.where('a == 1')

what about adding that for

expr.cast

as well. so instead of

daft.col('list').cast(daft.DataType.fixed_size_list(daft.DataType.int32(), 3))

you could just type:

daft.col('list').cast('int[3]')

ooo….

I guess the tricky thing might be how we construct that mapping of

str

:

DataType

? This should also just be consistent with our SQL representation of types right (edited)

I was thinking we'd just delegate directly to the sql engine like we do with

dataframe.where

Makes sense, and I guess we just have like a

DataType.from_sql("…")

? Could be useful

We didn't add it yet. It would be a great first PR for someone looking to get involved in daft sql!

Cool, is there an open issue yet? I can file one if not

@jay Does anyone else do this with their daft imports?

from daft import col, DataFrame, DataType as da

hmm I do find

DataType

quite verbose yeah

It’s mostly because DataFrame gets prioritized in intellisense, so you end up having to write most of it