Re: New feature request: FlashBack Query
От | Rod Taylor |
---|---|
Тема | Re: New feature request: FlashBack Query |
Дата | |
Msg-id | 74F771E7-C2B3-41F8-946A-AA7E83444EE3@gmail.com обсуждение исходный текст |
Ответ на | Re: New feature request: FlashBack Query ("Jonah H. Harris" <jonah.harris@gmail.com>) |
Ответы |
Re: New feature request: FlashBack Query
Re: New feature request: FlashBack Query |
Список | pgsql-hackers |
> > Wrong. When Oracle says it's committed, it's committed. No > difference between when, where, and how. In Oracle, the committed > version is *always* the first presented to the user... it takes time > to go back and look at older versions; but why shouldn't that be a bit > slower, it isn't common practice anyway. Same with rollbacks... why > should they optimize for them when 97% of transactions commit? Do 97% of transactions commit because Oracle has slow rollbacks and developers are working around that performance issue, or because they really commit? I have watched several developers that would prefer to issue numerous selects to verify things like foreign keys in the application in order to avoid a rollback. Anyway, I don't have experience with big Oracle applications but I'm not so sure that 97% of transactions would commit if rollbacks were cheaper.
В списке pgsql-hackers по дате отправления: