On Tue, Aug 8, 2017 at 2:25 PM, Alexander Korotkov <aekorotkov@gmail.com> wrote:
Do we already assume that default btree opclass for array element type matches PK opclass when using @>> operator on UPDATE/DELETE of referenced table?
I believe so, since it's a polymorphic function.
If so, we don't introduce additional restriction here...
You mean to remove the wrapper query ?
I think we should choose the query which would be better planned (and presumably faster executed). You can make some experiments and then choose the query.
GROUP BY would also use default btree/hash opclass for element type. It doesn't differ from DISTINCT from that point.