pg_dump future problem.

Поиск
Список
Период
Сортировка
От Christopher Kings-Lynne
Тема pg_dump future problem.
Дата
Msg-id 20030504120941.B43020-100000@houston.familyhealth.com.au
обсуждение исходный текст
Ответы Re: pg_dump future problem.  (Philip Warner <pjw@rhyme.com.au>)
Re: pg_dump future problem.  (Tom Lane <tgl@sss.pgh.pa.us>)
Список pgsql-hackers
I see an obvious problem in the way pg_dump dumps serials.  At the moment,
we have length 64 identifiers.  If someone has a 64 character table name
and adds a serial to it, it will get a truncated sequence name, with _seq
o the end.

pg_dump will dump it like this:

CREATE TABLE really_long_name (a SERIAL UNIQUE
);

SELECT SETVAL('really_long_na_seq', 120);

However, if we up 7.4, say, to use 128 character identifiers, then this
restore will fail, as the sequence name will NOT be truncated and the
setval() call will fail.

We really need:

ALTER SEQUENCE ON really_long_name(a) ....

Chris



В списке pgsql-hackers по дате отправления:

Предыдущее
От: "Jenny -"
Дата:
Сообщение: starting Postgresql's Postmaster
Следующее
От: Philip Warner
Дата:
Сообщение: Re: pg_dump future problem.