Peter Wilson <petew(at)yellowhawk(dot)co(dot)uk> writes:
> I found a while ago that after inserting a lot of rows into a clean
> Postgres table it would take several minutes just to analyse a command,
> not even starting the execution.
Oh? Could you provide a test case for this? I can certainly believe
that the planner might choose a bad plan if it has no statistics, but
it shouldn't take a long time to do it.
regards, tom lane