On Fri, Oct 22, 2021 at 11:59 AM Greg Nancarrow <gregn4...@gmail.com> wrote: > > On Fri, Oct 22, 2021 at 12:41 PM Greg Nancarrow <gregn4...@gmail.com> wrote: > > > > I was also previously concerned about what the behavior should be when > > only including just the partitions of a partitioned table in a > > publication using ALL TABLES IN SCHEMA and having > > publish_via_partition_root=true. It seems to implicitly include the > > partitioned table in the publication. But I did some testing and found > > that this is the current behavior when only the partitions are > > individually included in a publication using TABLE, so it seems to be > > OK. > > > > Thinking some more about this, it still may still be confusing to the > user if not explicitly stated in the ALL TABLES IN SCHEMA case. > How about adding some additional explanation to the end of the > following paragraph: > > + <para> > + When a partitioned table is published via schema level publication, all > + of its existing and future partitions irrespective of it being from the > + publication schema or not are implicitly considered to be part of the > + publication. So, even operations that are performed directly on a > + partition are also published via publications that its ancestors are > + part of. > + </para> > > Something like: > > Similarly, if a partition is published via schema level publication > and publish_via_partition_root=true, the parent partitioned table is > implicitly considered to be part of the publication, irrespective of > it being from the publication schema or not.
I have made this change in the v46 patch attached at [1]. [1] - https://www.postgresql.org/message-id/CALDaNm3kBrMO5EyEgK_TyOrBuw%2BRvdJ2mJfpWb5e8FbuKg2cQw%40mail.gmail.com Regards, Vignesh
