All non-equal predicates on a TimeUUID column are going to fail and fail
in silent way. The basic issue here is that when you use a comparison on
a time UUID column in C* it compares based on the Time portion of the UUID. When
Spark executes this filter (unhandled behavior) it will compare lexically, this
will lead to results being incorrectly filtered out of the set. As long as the
range predicate is handled completely by the connector the correct result
will be obtained.
All non-equal predicates on a TimeUUID column are going to fail and fail in silent way. The basic issue here is that when you use a comparison on a time UUID column in C* it compares based on the Time portion of the UUID. When Spark executes this filter (unhandled behavior) it will compare lexically, this will lead to results being incorrectly filtered out of the set. As long as the range predicate is handled completely by the connector the correct result will be obtained.