Sergey Konoplev writes:
> Is there a way (or workaround) to make it use i_test_table__created
> for frequent and i_test_table__tsvector_1 for sparse words?
You would need to update to 8.4 for that --- 8.3 does not have any
support for statistics estimation for @@.
(Whether 8.4 would get it right
Hi, All
Well what do we have?
8 core, 32 GB, RAID 10, CentOS 5.2, Pg 8.3
A query using tsearch in WHERE block with ORDER and LIMIT:
select * from test_table where obj_tsvector @@
make_tsquery('some_words', 'utf8_russian') and obj_status_did = 1
order by obj_created desc limit 10;
Two indexes -
