On Tue, Dec 04, 2018 at 03:41:59PM -0300, Alvaro Herrera wrote:
> I spent a few minutes trying to think of a way of determining which
> query to use at SQL-execution time -- two CTEs, one of which would be
> short-circuited ... but I couldn't figure out how. I also tried to use
> the new pg_partition_tree() function, but it's useless for this purpose
> because it roots at its argument table, and there doesn't seem to be a
> convenient way to obtain the topmost ancestor.
This has been mentioned on the thread where pg_partition_tree has been
discussed:
https://www.postgresql.org/message-id/6baeb45a-6222-6b88-342d-37fc7d3cf89a%40lab.ntt.co.jp
It got shaved from the final patch for simplicity as we had enough
issues to deal with first. Adding a pg_partition_root or a new column
in pg_partition_tree makes sense. My guts are telling me that a
separate function is more instinctive to use.
--
Michael