Closed as duplicate of#536
Description
Currently Arrow reading does not support nested types, such as for StructVector
or
ListVector
(which we don't support at all). We might need some recursive function, such as here for Maps in Spark to perform the type mapping, or at least let DataFrame infer the types like with a call to inferType()
.
I'm not sure whether Arrow allows for arbitrarily nested Lists.
But it seems like ListVector.field.children[].type
is available ;)
For writing, see #271