Re: fixing old_snapshot_threshold's time->xid mapping

Поиск
Список
Период
Сортировка
От Andres Freund
Тема Re: fixing old_snapshot_threshold's time->xid mapping
Дата
Msg-id 20200418201715.umih2jen27xmhxkn@alap3.anarazel.de
обсуждение исходный текст
Ответ на Re: fixing old_snapshot_threshold's time->xid mapping  (Thomas Munro <thomas.munro@gmail.com>)
Список pgsql-hackers
Hi,

On 2020-04-17 14:12:44 +1200, Thomas Munro wrote:
> What about a contrib function that lets you clobber
> oldSnapshotControl->current_timestamp?  It looks like all times in
> this system come ultimately from GetSnapshotCurrentTimestamp(), which
> uses that variable to make sure that time never goes backwards.

It'd be better than the current test situation, and probably would be
good to have as part of testing anyway (since it'd allow to make the
tests not take long / be racy on slow machines). But I still don't think
it really allows to test the feature in a natural way. It makes it
easier to test for know edge cases / problems, but not really discover
unknown ones. For that I think we need more granular bins.

- Andres



В списке pgsql-hackers по дате отправления:

Предыдущее
От: James Coleman
Дата:
Сообщение: Re: sqlsmith crash incremental sort
Следующее
От: Tom Lane
Дата:
Сообщение: Re: Poll: are people okay with function/operator table redesign?