On Fri, 2008-09-05 at 11:21 -0700, Tom Raney wrote:
> Why does the planner consider both input variations of each symmetric merge join? The README says "there is not a
lotof difference" between the two options. When are there any differences?
>
> -Tom Raney
>
http://archives.postgresql.org/pgsql-general/2008-08/msg00967.php
My understanding from that thread is that if one table has high
ndistinct and the other has low ndistinct, one plan may require more
re-scanning than the other.
Regards,Jeff Davis