The following bug has been logged online:

Bug reference: 2317
Logged by: TomasKlockar
Email address: tomask (AT) omicron (DOT) se
PostgreSQL version: 7.3.2/7.4.7
Operating system: linux(fedora)
Description: Wrong sorting order for (VW)
Details:

SELECT cname FROM clients ORDER BY cname;

sorts W before V and that is a mistake.

the result is

V
W
WHI
Vi
Wi
Volvo

Correct order would be to place all vV before all wW.

I think the database was initialized with UTF8 but it might have been
ISO_8859-1, however the error is easily repeatable and ill be happy to
provide a dump.

I tested this on 7.3.2 and 7.4.7 and the error was repeatable.

regards,

/Tomas

---------------------------(end of broadcast)---------------------------
TIP 6: explain analyze is your friend

On Mon, 13 Mar 2006, TomasKlockar wrote:

I think the most important thing we'd need to know is what locale the
database was initialized with in order to try to reproduce.

If you put similar data in a file and use the unix "sort" command with the
same locale, do you get the same order?


---------------------------(end of broadcast)---------------------------
TIP 6: explain analyze is your friend

"TomasKlockar" <tomask (AT) omicron (DOT) se> writes:
If the database encoding doesn't match the database locale (server
LC_COLLATE setting), all kinds of very strange sorting behavior can
ensue. I suspect you've been bit by this gotcha.

regards, tom lane

---------------------------(end of broadcast)---------------------------
TIP 2: Don't 'kill -9' the postmaster

That looks like the typical swedish locale, which sorts V and W as the
same character. Yes, that can be very annoying for some of us :-), but
that's the way it's defined.

//Magnus

---------------------------(end of broadcast)---------------------------
TIP 3: Have you checked our extensive FAQ?

http://www.postgresql.org/docs/faq

Well, PostgreSQL uses the locale functionatlity provided by your OS, so you=
will need to talk to them. Perhaps they already ship a different locale de=
finition that is more suitable for your needs that you can change to?=20

If you select locale=3DC you will get the sort all chars as separate charac=
ters. It will break your sorting of =E5=E4=F6, because they are not in the =
correct sequence in LATIN1 (or UTF8 for that matter), but you can certainly=
do that. It's a matter of picking the correct locale when you initdb your =
database.

//Magnus

---------------------------(end of broadcast)---------------------------
TIP 9: In versions below 8.0, the planner will ignore your desire to
choose an index scan if your joining column's datatypes do not
match

--wxDdMuZNg1r63Hyj
Content-Type: text/plain; charset=iso-8859-1
Content-Disposition: inline
Content-Transfer-Encoding: quoted-printable

On Thu, Mar 16, 2006 at 11:03:59AM +0100, Tomas Klockar wrote:
[...]
Tomas, as it has been said, PostgreSQL uses the operating system's
locale. So you can do two things: change it (e.b. to C , which is
neutral) or fix it. If you are on an unix-like OS, man 5 locale might
help with that.

Regards
-- tom=C3=A1s

--wxDdMuZNg1r63Hyj
Content-Type: application/pgp-signature; name="signature.asc"
Content-Description: Digital signature
Content-Disposition: inline

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.1 (GNU/Linux)

iD8DBQFEGk5bBcgs9XrR2kYRApU8AJ9qYVowY46ByjnCQwP1wL YrGsqtZACdHYYe
mERFVGkvj1cqPMQHBKv+Psc=
=sL7I
-----END PGP SIGNATURE-----

--wxDdMuZNg1r63Hyj--