I'm trying to find an efficient implementation of a percentile UDAF (or a
median or histogram function, since those would be built on the same
approach) for Impala 1.2/1.2.1. The Hive implementation is regrettably not
portable, as it's in Java (the Impala 1.2.1 docs claim Java UDAs are
acceptable as well as C++, but Impala actually throws an AnalysisExceptionw
ith "Java UDSas are not supported." if you try to actually add one).
Has anyone written one or have plans to release one?
To unsubscribe from this group and stop receiving emails from it, send an email to firstname.lastname@example.org.