Thanks both -- pushed. I also changed regress/sql/triggers to leave tables around that have a non-zero tgparentid. This ensures that the pg_upgrade test sees such objects, as well as findoidjoins.
I refrained from doing the findoidjoins dance itself, though; I got a large number of false positives that I think are caused by some pg12-era hacking. -- Álvaro Herrera https://www.2ndQuadrant.com/ PostgreSQL Development, 24x7 Support, Remote DBA, Training & Services
