Content-Type: text/plain;
  charset="iso-8859-1"
From: Nicholas Allen <nallen@freenet.co.uk>
To: Bruno Wolff III <bruno@wolff.to>
Subject: Re: How do you select from a table until a condition is met?
Date: Wed, 12 Feb 2003 20:55:21 +0100
User-Agent: KMail/1.4.1
Cc: Dmitry Tkach <dmitry@openratings.com>, pgsql-sql@postgresql.org
References: <200302121910.38232.nallen@freenet.co.uk>
	<200302122010.15297.nallen@freenet.co.uk>
	<20030212193538.GA29724@wolff.to>
In-Reply-To: <20030212193538.GA29724@wolff.to>
MIME-Version: 1.0
Content-Transfer-Encoding: quoted-printable
Message-Id: <200302122055.21790.nallen@freenet.co.uk>

I thought of this but the problem is that there may be multiple rows with t=
he=20
same value for the column I am sorting on. Eg if sorting on a surname then=
=20
there may be 100s of people with the same surname so generating a where=20
clause that selects up to the exact person previously selected is very=20
difficult.

eg where surname <=3D 'Jones' could return hundreds of records and probably=
 some=20
records for people with the same surname but which appear after the selecte=
d=20
record.

I had another idea though. Do you know if it is possible to write a functio=
n=20
for postgresql thatwill return true until a condition is met and then for=
=20
every row after that return false? This way I could write a function that=
=20
could select up to a particular row using the where clause and calling the=
=20
function from the where claus.

I'm sure there must be a way of doing this as it is such an obvious thing t=
o=20
need to do...

Thanks for the help!

On Wednesday 12 Feb 2003 8:35 pm, Bruno Wolff III wrote:
> On Wed, Feb 12, 2003 at 20:10:15 +0100,
>
>   Nicholas Allen <nallen@freenet.co.uk> wrote:
> > Thanks, but I don't think thats what I want. Basically I need to get the
> > row offset of a record (ie row) in a result set returned from the
> > database. I know all the details of the row in question but when the
> > search order changes I need to know the new index the row will have.
> >
> > In other words, I was hoping I would be able to select upto a particular
> > row (not knowing its offset but knowing everything else about it) in a
> > select query.
> >
> > Ideally I could do something like SELECT count(*) UNTIL col1=3Dvalue1 &&
> > col2=3Dval2 etc....
>
> Since you are ordering the data you can use where clauses with < (and/or =
>)
> operators to stop counting at the desired row. For this to work the
> ordering needs to be well defined. You can tack on the primary key to your
> list of columns to order by to make the ordering well defined if it isn't
> already.