olumn as a PK.
Cheers,
Matt
> -Original Message-
> From: Jeremy Zawodny [mailto:[EMAIL PROTECTED]
> Sent: 23 April 2004 23:51
> To: Emmett Bishop
> Cc: [EMAIL PROTECTED]
> Subject: Re: Compound Primary Key question
>
> On Fri, Apr 23, 2004 at 03:40:43PM -0700, E
On Fri, Apr 23, 2004 at 03:40:43PM -0700, Emmett Bishop wrote:
> Quick question. In general, is it better to create
> compound primary keys or use an auto increment field
> to uniquely identify each record?
Yes.
It depends on your application and your data.
Jeremy
--
Jeremy D. Zawodny | Pe
On Tue, Jan 22, 2002 at 10:19:46AM +0100, Roger Baklund wrote:
>
> Just to make it clear: this is the primary key, and all four parts
> of the index are needed to identify the row. Most of the time I use
> the three first parts, sometimes I use all four parts, and sometimes
> I use only the first
> Hi,
Hi, thank you for your reply.
> Since b can have only two values, making it the first column in the
> index columns will not be efficient,
Why?
> but since b is many distinct values
...I suppose you mean 'a'...
> ordering on this key first makes sense. Also you are always
> querying w
* Jeremy Zawodny
> On Mon, Jan 21, 2002 at 05:18:35PM +0100, Roger Baklund wrote:
> > Hi,
> >
> > Also, I wonder what is the best order for my primary keys:
> >
> > a intThis is the 'main' id, very many distinct values
> > b tinyintThis is a source code, possible values are 1 or 2
> >
On Mon, Jan 21, 2002 at 05:18:35PM +0100, Roger Baklund wrote:
> Hi,
>
> Also, I wonder what is the best order for my primary keys:
>
> a intThis is the 'main' id, very many distinct values
> b tinyintThis is a source code, possible values are 1 or 2
> c smallint Some counter
> d t
Hi,
A compound index creates an index in the order specified. If
key is (a,b,c,d) then the index is in the order a,b,cd. That is
the index file keeps these columns ordered with a pointer to
the actual row in the data table.
Since b can have only two values, making it the first column in the
in
