Hi,
I’m providing some cardinality estimates for JDBC tables. From the API, I can
easily implement a metadata handler for RowCount. But I’m encountering the
following exception:
RelSubset [rel#32:RelSubset#0.ENUMERABLE] has wrong best cost {5.999990709E8
rows, 5.999990719E8 cpu, 0.0 io}. Correct cost is {110.0 rows, 111.0 cpu, 0.0
io}
I’m assuming this is because the underlying RelOptTable has a default rowCount
of 100. Must I override this? Is it required to inherit JdbcTable to override
getStatistic() (and thus a custom JdbcSchema implementation to create these
instances)?
Thanks!
