Re: Pains in upgrading to 8.3

From: Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us>
To: Alvaro Herrera <alvherre(at)commandprompt(dot)com>
Cc: Bruce Momjian <bruce(at)momjian(dot)us>, Magnus Hagander <magnus(at)hagander(dot)net>, Dave Page <dpage(at)pgadmin(dot)org>, Tony Caduto <tony_caduto(at)amsoftwaredesign(dot)com>, paul rivers <rivers(dot)paul(at)gmail(dot)com>, Postgres General List <pgsql-general(at)postgresql(dot)org>
Subject: Re: Pains in upgrading to 8.3
Date: 2008-02-18 22:35:10
Message-ID: 11610.1203374110@sss.pgh.pa.us
Views: Raw Message | Whole Thread | Download mbox | Resend email
Thread:
Lists: pgsql-general

Alvaro Herrera <alvherre(at)commandprompt(dot)com> writes:
> Bruce Momjian escribi:
>> What cases on the past have needed the new pg_dump?

> Dependency handling IIRC in 7.3 (or was it 7.2?) was a big change for
> pg_dump, and I don't think we would have liked to backpatch the pg_dump
> changes. Also, AFAIK the sequences stuff with OWNED BY also needed the
> newer pg_dump, which is more recent (8.2?). I don't think it's as rare
> as you suggest.

Note that in straightforward cases, a dump from the older pg_dump will
serve perfectly well. However there have been cases in the past where
the dump/reload behavior of a PG version was unfixably broken for
certain corner cases --- renamed serial sequences being the latest
example. In such cases, dumping with the older pg_dump and loading into
a newer DB will not give you a correct copy of the state of your old
database, whereas dumping with the new pg_dump often does (because the
newer pg_dump knows how to exploit the feature we added to solve the
problem, such as OWNED BY).

The other usual reason for recommending this is the fear that someone's
old installation may not be up-to-date to its latest patch release, and
so its pg_dump might be missing bug fixes even for problems that *are*
soluble within its release series. The thought is that the version they
are installing probably is an up-to-date one, and so its pg_dump might
have fixes the other one doesn't.

None of this matters if you have a DB you can successfully dump and
reload in the old installation version. In that case you haven't got
any dump/reload bugs you are tripping over.

regards, tom lane

In response to

Browse pgsql-general by date

  From Date Subject
Next Message Reid Thompson 2008-02-19 00:09:27 Re: Auto incrementing primary keys
Previous Message Alvaro Herrera 2008-02-18 22:12:27 Re: SPI-functions and transaction control