Spotfire Ideas Portal
Statistics for data-source tables are independent of the query being executed, and while heuristics can be used to guess the volume of data retrieved & processed by different portions of a query plan, if -- for a configurable time-window -- the query engine retains cardinality information for fetches, joins, etc., then on subsequent invocations of the same query, TDV will be able to create plans that better reflect the actual data being processed.