rbalamohan
rbalamohan
Waiting till all inputs are ready is intensional and having this to timeout based approach could destabilize and cause corner case issues.
Tried it in internal cluster with 100 GB TPCH and shows good reduction in amount of bytes read from s3. select count(*) from customer where c_custkey > 14099999 and c_custkey...
compute_bit_vector is vectorized via https://issues.apache.org/jira/browse/HIVE-24510
LGTM. +1. BTW, pipelinesorter is the brainchild of @t3rmin4t0r . :)
