On Wed, Jan 11, 2023 at 3:00 PM Andres Freund <and...@anarazel.de> wrote: > What are your thoughts about the place for the helper functions? You're ok > with rmgrdesc_utils.[ch]?
Yeah, that seems okay. We may well need to put more stuff in that file. We're overdue a big overhaul of the rmgr output, so that everybody uses the same format for everything. We made some progress on that for 16 already, by standardizing on the name snapshotConflictHorizon, but a lot of annoying inconsistencies still remain. Like the punctuation issue you mentioned. Ideally we'd be able to make the output more easy to manipulate via the SQL interface from pg_walinspect, or perhaps via scripting. That would require some rules that are imposed top-down, so that consumers of the data can make certain general assumptions. But that's fairly natural. It's not like there is just inherently a great deal of diversity that we need to be considered. For example, the WAL records used by each individual index access method are all very similar. In fact the most important index AM WAL records used by each index AM (e.g. insert, delete, vacuum) have virtually the same format as each other already. -- Peter Geoghegan