On Thu, Aug 26, 2021 at 11:10 PM Ilya Gladyshev <i.gladys...@postgrespro.ru> wrote: > I have developed a simple patch to fix this, while I’m not fully satisfied > with it, it gets the job done.
Thanks for working on this! > In message [1] there’s also mentioned possibility of existence of triggers on > the foreign server, and transforming the update to delete+insert will cause > these triggers to be omitted. Yeah, I still think so. > While it is true, I feel like partition pruning already has a similar effect, > as it allows to skip scanning foreign partitions. I don’t fully understand this part. Could you elaborate a bit more on it? > The only way to both fix the update and have the triggers work, that I came > up with, is to use parent partitioned table as a target for the foreign > update (FDW request would then be "UPDATE public.players …"), while this is > possible, it requires the foreign server to have the same partitioned table, > which seems quite restrictive to me. Yeah, that would be too restrictive. Best regards, Etsuro Fujita
