Subject: Re: Atomicity of UPDATE, interchanging values in unique
From: Rod Taylor <rbt@rbt.ca>
To: daniel alvarez <d-alvarez@gmx.de>
Cc: pgsql-sql@postgresql.org
In-Reply-To: <21780.1047160126@www48.gmx.net>
References: <1047159202.28251.196.camel@jester>
	<21780.1047160126@www48.gmx.net>
Content-Type: multipart/signed; micalg=pgp-sha1;
	protocol="application/pgp-signature";
	boundary="=-CDGK5ZDmmAgloZZQvzhY"
Organization: 
Message-Id: <1047160571.28251.202.camel@jester>
Mime-Version: 1.0
Date: 08 Mar 2003 16:56:11 -0500

--=-CDGK5ZDmmAgloZZQvzhY
Content-Type: text/plain
Content-Transfer-Encoding: quoted-printable

On Sat, 2003-03-08 at 16:48, daniel alvarez wrote:
> > The first is what you want.  PostgreSQL needs some work in the
> > evaluation of unique indexes to properly support it.
> >=20
> > Namely, when it sees a conflict when inserting into the index, it needs
> > to record the fact, and revisit the conflict at the end of the command.=
=20
> > Lots of work...
>=20
> OK. The long-term goal would then be to get rid of such oddities. But what
> can I do right now as a user to solve that issue for my application?

Certainly.. But you have to find someone willing to do a the work for
little gain. There are lots of issues more important to most of the
developers.

> There must be a better solution than the additional dummy update.

You could try hiding it behind a function, but I'm afraid thats the only
sane way to do it.  Select into temp table, delete both, and insert
values back in again is another :)

--=20
Rod Taylor <rbt@rbt.ca>

PGP Key: http://www.rbt.ca/rbtpub.asc

--=-CDGK5ZDmmAgloZZQvzhY
Content-Type: application/pgp-signature; name=signature.asc
Content-Description: This is a digitally signed message part

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.2.1 (FreeBSD)

iD8DBQA+amb76DETLow6vwwRAmtUAJ9cIunnMSy5Zd5FJgzWFA0bJmGfAACeNcO/
rVvFn1VuQjICRf0LMsBn4VE=
=XWXc
-----END PGP SIGNATURE-----

--=-CDGK5ZDmmAgloZZQvzhY--