LWN: Comments on "Declarative partitioning in PostgreSQL"

Declarative partitioning in PostgreSQL

daroc — Fri, 12 Apr 2024 13:06:41 +0000

I don't believe partitions can span separate hosts; although I don't recall the question coming up at the time, so perhaps I am mistaken. I think the fundamental problem would be having some way to introduce consistent constraints across separate instances of PostgreSQL, which sounds hard.

Declarative partitioning in PostgreSQL

grifferz — Thu, 11 Apr 2024 22:18:38 +0000

With either the old or the new method, can partitioned tables span multiple hosts, or does that kind of partitioning still have to be done entirely in the client logic?

Declarative partitioning in PostgreSQL

mirabilos — Fri, 29 Mar 2024 06:28:10 +0000

How can this be called nearly feature-complete when whole-partitioned-table indexes (necessary for constraints) are still not even on the horizon?

As to the names… ouch, yes, hadn’t thought about that. Is it possible to make it so all child tables go into a separate schema (one schema per parent table) as kind of namespacing?