Re: mvcc catalo gsnapshots and TopTransactionContext
От | Robert Haas |
---|---|
Тема | Re: mvcc catalo gsnapshots and TopTransactionContext |
Дата | |
Msg-id | CA+TgmoZcAGJd09pWeNg6mYd5VM373=gqMuZOtgmA_5sL0LeqnA@mail.gmail.com обсуждение исходный текст |
Ответ на | Re: mvcc catalo gsnapshots and TopTransactionContext (Andres Freund <andres@2ndquadrant.com>) |
Ответы |
Re: mvcc catalo gsnapshots and TopTransactionContext
|
Список | pgsql-hackers |
On Tue, Aug 6, 2013 at 3:06 AM, Andres Freund <andres@2ndquadrant.com> wrote: > The reason for that is that when we do the AtEO(Sub)?Xact_Inval(), we've > already done a RecordTransactionAbort(true|false) and > CurrentTransactionState->state = TRANS_ABORT. So the visibility routines > have enough information to consider rows created by the aborted > transaction as invisible. > > I am not really happy with the RelationReloadIndexInfo()s in > RelationClearRelation() when we're in an aborted state, especially as > the comment surrounding them are clearly out of date, but I don't see a > bug there anymore. How can it be safe to try to read catalogs if the transaction is aborted? -- Robert Haas EnterpriseDB: http://www.enterprisedb.com The Enterprise PostgreSQL Company
В списке pgsql-hackers по дате отправления: