hi,
why not using "primary key (a , b) " clause at the end of your create table
statement?It would unsure you that "b" is unique for particular "a". I
suggest that you test the null value entries... if you are going to face
with.
regards,
omid.
>From: Jose Soares <jose@sferacarta.com>
>To: Lincoln Yeoh <lylyeoh@mecomb.com>
>CC: "pgsql-general@postgreSQL.org" <pgsql-general@postgreSQL.org>
>Subject: Re: [GENERAL] select for insert?
>Date: Tue, 18 Jan 2000 14:41:34 +0100
>
>
>
>Lincoln Yeoh wrote:
>
> > Hi everyone,
> >
> > say I have a table test
> > create table test (a int, b text);
> >
> > and I want to ensure that 'b' is unique for a particular 'a' ('a' not
>being
> > unique).
> >
> > Right now I can do something like
> > begin transaction
> > select count(*) from test where b =$value;
> > if count = 0 then insert into test (a,b) values ($a, $value);
> >
> > But there is still a chance of a duplicate being inserted.
>
>use unique constraint on column a as in:
>
>CREATE TABLE test (a char unique, b int);
>
>Jos�
>
>
>
>************
>
______________________________________________________
Get Your Private, Free Email at http://www.hotmail.com
