Re: hashagg slowdown due to spill changes

Поиск
Список
Период
Сортировка
От Melanie Plageman
Тема Re: hashagg slowdown due to spill changes
Дата
Msg-id CAAKRu_ZVD47byoEdR7C=nv2mf3SYB+qqQVmHJi4ON0b+PhvxqA@mail.gmail.com
обсуждение исходный текст
Ответ на Re: hashagg slowdown due to spill changes  (Andres Freund <andres@anarazel.de>)
Ответы Re: hashagg slowdown due to spill changes  (Andres Freund <andres@anarazel.de>)
Список pgsql-hackers

On Mon, Jun 22, 2020 at 9:02 PM Andres Freund <andres@anarazel.de> wrote:

>  /*
> - * find_unaggregated_cols
> - *     Construct a bitmapset of the column numbers of un-aggregated Vars
> - *     appearing in our targetlist and qual (HAVING clause)
> + * Walk tlist and qual to find referenced colnos, dividing them into
> + * aggregated and unaggregated sets.
>   */
> -static Bitmapset *
> -find_unaggregated_cols(AggState *aggstate)
> +static void
> +find_cols(AggState *aggstate, Bitmapset **aggregated, Bitmapset **unaggregated)
>  {

It's not this patch's fault, but none, really none, of this stuff should
be in the executor.


Were you thinking it could be done in grouping_planner() and then the
bitmaps could be saved in the PlannedStmt?
Or would you have to wait until query_planner()? Or are you imagining
somewhere else entirely?

--
Melanie Plageman

В списке pgsql-hackers по дате отправления:

Предыдущее
От: Alvaro Herrera
Дата:
Сообщение: Re: pg_resetwal --next-transaction-id may cause database failed torestart.
Следующее
От: Andres Freund
Дата:
Сообщение: Re: hashagg slowdown due to spill changes