On 15 Září 2011, 16:26, Tom Lane wrote: > Yang Zhang <yanghatespam@xxxxxxxxx> writes: >> On Thu, Sep 15, 2011 at 1:22 AM, Tomas Vondra <tv@xxxxxxxx> wrote: >>> [ use a left join instead of NOT IN ] > >> This worked great, thank you. Too bad the planner isn't smart enough >> to do this yet! > > It never will be, because they're not equivalent queries. NOT IN has > different (and very hard-to-optimize) behavior in the face of NULLs in > the right-hand table. True, but in this case the subquery selects a PK column (at least I believe so, we haven't seen the structure), so there should be no NULLs. So the result should be the same. Could the planner do this reasoning, i.e. see if there's a NOT NULL constraint on the column and change the plan accordingly? Tomas -- Sent via pgsql-general mailing list (pgsql-general@xxxxxxxxxxxxxx) To make changes to your subscription: http://www.postgresql.org/mailpref/pgsql-general