Thomas Munro <> writes:
> As noted in a nearby review of some similar code[1], commit 7a2fe9bd
> made merge append very slightly more efficient, but nobody told
> cost_merge_append about the change.  I doubt it makes much difference
> to the final cost in practice but I figured it might be worth
> correcting the comment.  Does this make sense?

Agreed, pushed.

                        regards, tom lane

Sent via pgsql-hackers mailing list (
To make changes to your subscription:

Reply via email to