On Tue, 2009-01-27 at 04:28 +0100, marcin mank wrote:
> On Mon, Jan 26, 2009 at 10:37 PM, Dann Corbit <DCorbit@connx.com> wrote:
> > My notion is to do a character mode database dump as SQL statements and
> > then load into the new version by execution of psql against the sql
> > STATEMENTS.
> >
> > What are the "gotchas" we can expect with this approach?
> >
> > When I say 'ancient' I mean v7.1.3 and the target is v8.3.5.
> >
>
> One gotcha is that some ancient versions (and I think 7.1 is one of
> these) used to silently truncate varchar values that don`t fit into
> the declared field width, while recent versions throw errors. Check if
> Your apps don`t depend on this behaviour.'
Wow that is reaching back. You can actually do this still:
postgres=# create table test_trunc(fname varchar(2)):
postgres-#
postgres=# create table test_trunc(fname varchar(2));
CREATE TABLE
postgres=# insert into test_trunc values ('fo');
INSERT 0 1
postgres=# insert into test_trunc values ('foo');
ERROR: value too long for type character varying(2)
postgres=# insert into test_trunc values ('foo'::varchar(2));
INSERT 0 1
postgres=# select * from test_trunc;
fname
-------
fo
fo
(2 rows)
Sincerely,
Joshua D. Drake
> good luck
> Marcin
>
--
PostgreSQL - XMPP: jdrake@jabber.postgresql.org
Consulting, Development, Support, Training
503-667-4564 - http://www.commandprompt.com/
The PostgreSQL Company, serving since 1997
