On Mon, Oct 7, 2019 at 5:56 PM Ron <ronljohnso...@gmail.com> wrote: > On 10/7/19 6:17 PM, Michael Lewis wrote: > > No, what you want is not possible and probably won't ever be I would > expect. > > Sure it is. Maybe not the (weird) way that Postgres does partitioning, > but > the legacy RDBMS that I still occasionally maintain has for at least 25 > years had partition key independent of any indexes. > > > Scanning every partition to validate the primary key isn't scalable. > > That's only because of the way Pg implements partitioning. >
I can dig that, but since this is a Postgres list and everything I have heard indicates it is not a limitation that is likely to be removed in Postgres, it seems like we are having two different discussions.