Hi everyone,
Referring to the regress test feedback from CFBot against the V5 patch, I have reverted
the modifications made to `btcostestimate()`, and revised `genericcostestimate()` as follows:
When `numIndexTuples` gets recalculated within this function, `indexSelectivity` will be
re-estimated based on `index->rel->tuples`, ensuring that `numIndexTuples` and `indexSelectivity`
are mutually consistent.
I have verified the fix against the previously failing regression tests listed by CFBot,
specifically `generated_stored`, `generated_virtual`, `subselect` and `select_parallel`,
and all of them now pass successfully.
regards,
--
ZizhuanLiu (X-MAN)
44973863@qq.com
