On Thu, Dec 06, 2018 at 10:48:59PM -0300, Alvaro Herrera wrote:
> I think adding a pg_partition_root() call to as many pg_partition_tree
> tests as you modified is overkill ... OTOH I'd have one test that
> invokes pg_partition_tree(pg_partition_root(some-partition)) to verify
> that starting from any point in the tree you get the whole tree.
Good idea, thanks for the input.
> I haven't actually tried to write a query that obtains a tree of
> constraints using this, mind ...
Sure. It would be good to agree on an interface. I have not tried
either, but you should be able to get away with a join on relid returned
by pg_partition_tree() with pg_constraint.conrelid with
pg_get_constraintdef() instead of a WITH RECURSIVE, no?
--
Michael