Steven Schlansker <ste...@likeness.com> writes: > It's been covered a few times in the past, > http://www.postgresql.org/message-id/banlktimft4ohqkb6y7m4wqrffpwnutp...@mail.gmail.com > http://postgresql.1045698.n5.nabble.com/partial-indexes-not-used-on-parameterized-queries-td2121027.html > but in a nutshell, partial indices do not play nicely with prepared > statements because whether the index is valid or not cannot be known > at query plan time.
This should be pretty much a non-issue in 9.2 and up; if the partial index is actually useful enough to be worth worrying about, the plancache choice logic will realize that it should use custom not generic plans. http://git.postgresql.org/gitweb/?p=postgresql.git&a=commitdiff&h=e6faf910d You might still have an issue if the partial index is only sometimes usable --- the choice logic might decide to go for the generic-plan approach anyway. But if you've got a case where the optimal plan is all over the map like that, I wonder why you're using a prepared statement at all ... regards, tom lane -- Sent via pgsql-general mailing list (pgsql-general@postgresql.org) To make changes to your subscription: http://www.postgresql.org/mailpref/pgsql-general
