Horton

Re: Show various offset arrays for heap WAL records

Peter Geoghegan <pg@bowt.ie>

View thread View on postgr.es
From: Peter Geoghegan <pg@bowt.ie>
To: Melanie Plageman <melanieplageman@gmail.com>
Cc: Andres Freund <andres@anarazel.de>, pgsql-hackers@postgresql.org, Robert Haas <robertmhaas@gmail.com>, Peter Eisentraut <peter.eisentraut@enterprisedb.com>
Date: 2023-04-11T19:22:12Z
Lists: pgsql-hackers

Attachments

On Tue, Apr 11, 2023 at 11:48 AM Peter Geoghegan <pg@bowt.ie> wrote:
> Attached revision deals with this by spelling out the names in full
> (e.g., "old_xmax" and "new_xmax"). It also reorders the output fields
> to match the order from the physical UPDATE, HOT_UPDATE, and LOCK WAL
> record types, on the theory that those should match the physical
> record (unless there is a good reason not to, which doesn't apply
> here).

I just noticed that we don't even show xmax in the case of DELETE
records. Perhaps the original assumption is that it must match the
record's own XID, but that's not true after the MultiXact enhancements
for foreign key locking added to 9.3 (and in any case there is no
reason at all to show xmax in UPDATE but not in DELETE).

Attached revision v4 fixes this, making DELETE, UPDATE, HOT_UPDATE,
LOCK, and LOCK_UPDATED record types consistent with each other in
terms of the key names output by the heap desc routine. The field
order also needed a couple of tweaks for struct consistency (and
cross-record consistency) for v4.

-- 
Peter Geoghegan

Commits

  1. Merge prune, freeze and vacuum WAL record formats

  2. Add rmgrdesc README

  3. Refine the guidelines for rmgrdesc authors.

  4. Fix Heap rmgr's desc output for infobits arrays.

  5. Clarify nbtree posting list update desc issue.

  6. Fix nbtree posting list update desc output.

  7. Set cutoff xmin more aggressively when vacuuming a temporary table.