nullable in spark-sql

[Koert Kuipers]

In spark 2, is nullable treated as reliable? or is it just a hint for
efficient code generation, the optimizer etc.

The reason i ask is i see a lot of code generated with if statements
handling null for struct fields where nullable=false