On Sun, 30 Jun 2019 at 17:54, David Rowley <david.rowley@2ndquadrant.com> wrote:
> Any further thoughts on this Michael?
>
> Or Andres? Do you have a preference to which of the approaches
> (mentioned upthread) I use for the fix?
>
> If I don't hear anything I'll probably just push the first fix. The
> inefficiency does not affect heap, so likely the people with the most
> interest in improving that will be authors of other table AMs that
> actually do something during table_finish_bulk_insert() for
> partitions. We could revisit this in PG13 if someone comes up with a
> need to improve things here.
I've pushed the original patch plus a small change to only call
table_finish_bulk_insert() for the target of the copy when we're using
bulk inserts.
--
David Rowley http://www.2ndQuadrant.com/
PostgreSQL Development, 24x7 Support, Training & Services