> On Aug 25, 2022, at 21:43, Bryn Llewellyn <bryn@xxxxxxxxxxxx> wrote: > [...] I've read this a few times, and I am having trouble understanding what behavior you were expecting out of PostgreSQL, and what behavior you received that you didn't think was correct. If it is "pg_constraint has a column connamespace, and that appears to be a denormalization since a constraint is always in the same schema as the table it is owned by," I believe Tom explained the reason for that. If that's not what is concerning you, can you summarize it in a sentence two?