Re: Reducing the WAL overhead of freezing in VACUUM by deduplicating per-tuple freeze plans

On Thu, Nov 10, 2022 at 7:00 PM Justin Pryzby <pryzby(at)telsasoft(dot)com> wrote:
> Note that this comment is dangling in your patch:

Attached is v4, which removes the old comments you pointed out were
now out of place (they weren't adding much anyway). Also fixed bitrot
against HEAD from today's visibility map commit from Jeff Davis.

There is a more substantive change here, too. Like v3, v4 refactors
the *mechanical* details of how the XID based cutoff is handed down.
However, unlike v3, v4 goes back to using vacuumlazy.c's FreezeLimit
as the starting point for generating a latestRemovedXid. It seemed
better to deal with the recovery conflict issues created by my big
page-level freezing/freezing strategy patch set in the patch set
itself.

Will commit this early next week barring any objections.

Thanks
--
Peter Geoghegan