Execute over dicts - nullability

[amachanic]

Couple of questions regarding executing over dicts:

A) Am I correct in assuming that Python None == SQL NULL? If so, can sqlglot currently support aggregation including NULLs, without throwing an unsupported operand type exception?
B) Any thoughts on potentially supporting sparse dicts? E.g. [{'x': 0, 'y': 1}, {'x': 1}]. If both 'x' and 'y' are defined in a schema, then 'y' in the second dict could be treated as NULL.
C) Along the same lines as (B), is it possible in a schema to define nullability?

Thanks!

By the way, this thing is really cool. Great job. I think this will be super useful for ad hoc data exploration and engineering.

[tobymao]

Hey @amachanic, First and foremost, the execution engine is really early and not really ready for any real use cases.

But, I'm happy to have a discussion.

A) Yes right now Python None == SQL Null. I think right now it would error out because i don't handle nulls very well. Again, the engine is super early.

B) it could make type inference a bit harder, but it could happen

c) we have a data type called Nullable to define nullability but it isn't really used anywhere

[tobymao]

nullability is now handled