Thomas Munro writes:
> While looking at bug #15857[1], I wondered why the following two
> queries get different plans, given the schema and data from the bug
> report:
> ...
> Here's my question: how could it ever be OK to sort/unique something
> and put it in a hash table, but not OK to put exact
Hello,
While looking at bug #15857[1], I wondered why the following two
queries get different plans, given the schema and data from the bug
report:
(1) SELECT COUNT (*)
FROM a
JOIN b
ON a.id=b.base_id
WHERE EXISTS (
SELECT 1
FROM c
WHER
