On Mon, May 20, 2019 at 09:42:52PM -0700, Paul A Jungwirth wrote:
> I noticed the docs at
> https://www.postgresql.org/docs/devel/ddl-partitioning.html still say
> you can't create a foreign key referencing a partitioned table, even
> though the docs for
> https://www.postgresql.org/docs/devel/sql-createtable.html have been
> updated (compared to v11). My understanding is that foreign keys
> *still* don't work as expected when pointing at traditional INHERITS
> tables, but they *will* work with declaratively-partitioned tables. In
> that case I suggest this change:
Could you define what is an "inheritance-partitioned" table? I know
of partitioned tables, inherited tables and tables which make use
of inheritance for partitioning (hence Inheritance Partitioning), but
the paragraph you are adding introduces a new term in the whole tree.
This makes things confusing and the previous paragraph is not, even if
it is incorrect since f56f8f8 as Amit has noted.
A simple rewording could be "tables using inheritance for
partitioning".
> (I've also attached it as a patch file.) In other words, we should
> move this caveat from the section on declaratively-partitioned tables
> to the section on inheritance-partitioned tables.
As this restriction only applies only to tables partitioned with
inheritance, then this move should be fine.
--
Michael