On Thu, 20 May 2021 at 12:20, tsunakawa.ta...@fujitsu.com <tsunakawa.ta...@fujitsu.com> wrote: > Yes, I want to make/keep it possible that application developers can be > unaware of partitions. I believe that's why David-san, Alvaro-san, and you > have made great efforts to improve partitioning performance. So, I'm +1 for > what Hou-san is trying to achieve. > > Is there something you're concerned about? The amount and/or complexity of > added code?
It would be good to see how close Amit's patch gets to the performance of the original patch on this thread. As far as I can see, the difference is, aside from the setup code to determine if the partition is constant, that Amit's patch just requires an additional ExecPartitionCheck() call per row. That should be pretty cheap when compared to the binary search to find the partition for a RANGE or LIST partitioned table. Houzj didn't mention how the table in the test was partitioned, so it's hard to speculate how many comparisons would be done during a binary search. Or maybe it was HASH partitioned and there was no binary search. David
