Thanks for references, Justin!
Couple comments about these references.
1) "Make unlogged table resets detectable".
https://www.postgresql.org/message-id/flat/62750df5b126e1d8ee039a79ef3cc64ac3d47cd5.camel%40j-davis.com
This conversation is about specific problem (unlogged table repairing).
Operation log has another use - it is primary a helper for diagnostics.
2) "RFC: Add 'taint' field to pg_control."
https://www.postgresql.org/message-id/flat/20180228214311.jclah37cwh572t2c%40alap3.anarazel.de
This is almost the same problem that we want to solve with operation
log. Differences between the operation log and what is discussed in the
thread:
* there suggested to store operation log in pg_control file - but
separate from pg_control main data (and write data separately too);
* operation log data can be represented in relational form (not flags),
this is more usable for RDBMS;
* number of registered event types can be increased easy (without
changes of storage).
--
With best regards,
Dmitry Koval
Postgres Professional: http://postgrespro.com