Hello, I'm working on a custom storage handler for Cassandra that was built in HIVE 0.11 and run on SPARK 0.9.2 to HIVE 1.2.1 and SPARK 1.6. I appreciate your help with three questions:
1) Hive 0.11 populated a parameter in the jobConf, TableScanDesc.FILTER_TEXT_CONF_STR, and it appears to me that HIVE on SPARK does not do this. Is this confirmed? We are implementing the DecomposedPredicate Interface, but it does not seem to be called and our ParsedPredicate relied on the FILTER information. Sarg.pushdown is also empty. I was able to find some references to SPARK not doing this with patches to support parquet, but is there a universal workaround? 2) We wrote the storage handler to operate as an External Table in Hive. I know that SPARK also would support a C* storage handler, but thought that we could continue to use our storage handler at Hive's level. Is this the case? 3) Do you have any further advice for this upgrade? Thank you, ~ Shawn M Lavelle
