(2018/07/09 20:06), Ashutosh Bapat wrote:
> On Mon, Jul 9, 2018 at 4:33 PM, Etsuro Fujita
> <fujita.etsuro@lab.ntt.co.jp> wrote:
>>> As I said, we do spend cycles in that function testing whether a node
>>> is Aggref or not even when the query doesn't have aggregates or
>>> grouping OR spend cycles in testing whether a node is a PlaceHolderVar
>>> when the query doesn't produce any. So, I don't see any problem with
>>> spending a few cycles testing whether a node is ConvertRowtypeExpr or
>>> not when a ConvertRowtypeExpr is not in the query or command. That's
>>> not a huge performance trouble. I would be happy to change my mind, if
>>> you show me performance different with and without this patch in
>>> planning. I haven't seen any.
>>
>>
>> I have to admit that the case in [1] wouldn't affect the performance, but my
>> concern is that there might be some cases where the test affects
>> performance.
>
> What are those cases? Can you please provide any numbers supporting your claim?
I don't have any numbers right now, so that is nothing but a concern.
But as I said in a previous email, in the approach I proposed, we don't
need to spend extra cycles where partitioning is not involved. I think
that is a good thing in itself. No?
Best regards,
Etsuro Fujita