On Sat, Mar 06, 2021 at 06:50:13PM +0900, Kohei KaiGai wrote:
>
> On the other hand, I'm uncertain whether the pfree(new_path) at the tail
> of add_path makes sense on the modern hardware, because they allow to
> recycle just small amount of memory, then entire memory consumption
> by the optimizer shall be released by MemoryContext mechanism.
> If add_path does not release path-node, the above portion is much simpler.
>
Hi Kaigai-san,
Do you have an updated patch? Please feel free to resubmit to next CF.
Current CF entry has been marked as RwF.
--
Jaime Casanova
Director de Servicios Profesionales
SystemGuards - Consultores de PostgreSQL