On Sun, 2009-08-16 at 18:32 -0700, Jeff Davis wrote:
> If VACUUM freezes one tuple on a page, it's likely that there are others
> on the same page that are close to vacuum_freeze_min_age, but not quite.
> Because the page is already dirty from freezing one tuple, it makes
> sense to be more aggressive about freezing the rest, in the hope that
> all the tuples will be frozen, and we will not have to dirty the page
> again later.
In the old days, where all new tuples were put at the end, this would
have made a lot of sense. But nowadays, with fillfacter, HOT, and so
on, it's quite likely that all the stuff around an outdated tuple are
newer versions of the same tuple or newer versions of other tuples close
by.
The patch might make sense anyway, but I think it might not be such an
overwhelming winner in practice.