Re: arbitrary number of values from a sequence
От | Gyozo Papp |
---|---|
Тема | Re: arbitrary number of values from a sequence |
Дата | |
Msg-id | 005d01c0d55f$932d41c0$4647c5d5@jaguar обсуждение исходный текст |
Ответ на | arbitrary number of values from a sequence ("Gyozo Papp" <pgerzson@freestart.hu>) |
Ответы |
Re: arbitrary number of values from a sequence
|
Список | pgsql-general |
Ok. I'm looking for another solution. The reason why I'm not dealing with sequence's increment is that there is no way to set an appropiate limit, sometimes I need 5, sometimes 17. Thanks for your help, Papp Gyozo - pgerzson@freestart.hu ----- Original Message ----- From: "Tom Lane" <tgl@sss.pgh.pa.us> To: "Gyozo Papp" <pgerzson@freestart.hu> Cc: <pgsql-general@postgresql.org> Sent: 2001. május 4. 20:26 Subject: Re: [GENERAL] arbitrary number of values from a sequence > "Gyozo Papp" <pgerzson@freestart.hu> writes: > > Does it mean that it can't be ensured that returning values of > > nextval() are consecutive ones? > > I think it would be folly to assume that, even with a cache setting > equal to the number of values you intend to fetch. If the cache gets > out of sync with your requests (say, because a transaction aborted after > fetching just some of the 5 values) then subsequent transactions would > reload the cache partway through, and in that case you could get > non-consecutive results. > > > does it help me if I set the transaction isolation level to > > serializable or lock the table of the sequence? > > No. Why do you need such a thing, anyway? If you are always allocating > groups of 5 IDs, why don't you just pretend each nextval() gives you > five items instead of one? You could simply multiply the returned value > by 5. Or set the sequence's increment to 5. > > regards, tom lane
В списке pgsql-general по дате отправления: