Github user lokm01 commented on the issue:
https://github.com/apache/spark/pull/21215
@maropu Thanks! Didn't know about creating a literal this way.
Don't you feel that the suggested change is way more elegant
Github user lokm01 commented on the issue:
https://github.com/apache/spark/pull/21215
@maropu That would work if you had scala case classes for all the types. In
our case, we're working on a generic framework, where we only have Spark
schemas (and I'd rather not generate case classes
Github user lokm01 commented on the issue:
https://github.com/apache/spark/pull/21215
Hey @maropu,
So we've encountered a number of issues with casting:
1. Casting an empty array to an array of primitive types caused an
exception on 2.2.1, but works on 2.3.0+ so that's
Github user lokm01 commented on the issue:
https://github.com/apache/spark/pull/21121
@ueshin Currently we use our own implementation of zipWithIndex when we do
explode and need to preserve the ordering of the array elements (especially if
there is a shuffle involved