Re: postgres_fdw batching vs. (re)creating the tuple slots

Поиск
Список
Период
Сортировка
От Tom Lane
Тема Re: postgres_fdw batching vs. (re)creating the tuple slots
Дата
Msg-id 2433405.1622416631@sss.pgh.pa.us
обсуждение исходный текст
Ответ на Re: postgres_fdw batching vs. (re)creating the tuple slots  (Andres Freund <andres@anarazel.de>)
Ответы Re: postgres_fdw batching vs. (re)creating the tuple slots  (Tomas Vondra <tomas.vondra@enterprisedb.com>)
Список pgsql-hackers
Andres Freund <andres@anarazel.de> writes:
> On 2021-05-30 17:10:59 -0400, Tom Lane wrote:
>> But it does seem like the hashing scheme somebody added to resowners
>> is a bit too simplistic.  It ought to be able to cope with lots of
>> refs to the same object, or at least not be extra-awful for that case.

> It's not really the hashing that's the problem, right? The array
> representation would have nearly the same problem, I think?

ResourceArrayAdd would have zero problem.  ResourceArrayRemove is
O(1) as long as resources are removed in reverse order ... which
is effectively true if they're all the same resource.  So while
I've not tested, I believe that this particular case would have
no issue at all with the old resowner implementation, stupid
though that was.

> It doesn't seem trivial to improve it without making resowner.c's
> representation a good bit more complicated.

Dunno, I have not studied the new version at all.

            regards, tom lane



В списке pgsql-hackers по дате отправления:

Предыдущее
От: Thomas Munro
Дата:
Сообщение: Re: O_DIRECT on macOS
Следующее
От: Michael Paquier
Дата:
Сообщение: Re: be-secure-gssapi.c and auth.c with setenv() not compatible on Windows