David Welton <davidnwelton@xxxxxxxxx> writes: > I'm curious as to what the general opinion is on these different > queries and their relative merits, especially in terms of speed: You really should do EXPLAIN ANALYZE rather than assuming that the planner's estimates are reliable. The NOT IN case, at least, could be improved by increasing work_mem to the point where the subquery fits into an in-memory hash table. Larger work_mem would probably help the sorts used in the other plans too. regards, tom lane
