Re: SQL Question

----- Original Message -----
From: "R. Bruce Hoffman, Jr." <rbruceh@attglobal.net>
To: <MIDRANGE-L@midrange.com>
Sent: Thursday, June 29, 2000 18:23
Subject: Re: SQL Question


>Select a.OrderNo, a.Customer, b.Seqno, b.Product
>From Orders a Join Items b
>Where b.Seqno =
>        ( Select Min(c.Seqno) From Items c
>              Where c.OrderNo = a.OrderNo
>              Group by c.OrderNo )
>

> sidebar...  you don't (shouldn't) need the the group by clause in a
> correlated subquery as the selection is already limited to a single order
> number.  In V4R2 (definitely) and up you don't.  The requirement here is
> that the correlated subquery return a single value, and with just the
where
> clause, it does so.

You're absolutely correct Bruce. It isn't a requirement.

Just keep in mind that on the AS/400, the access plan is determined by the
query optimizer. In the above example,  the GROUP BY clause causes the
optimizer to utilize a processing method referred to as "Index Skip Key
processing". This improves performance by eliminating the need to process
all index key values for the group, if there is an appropriate index built
over the (OrderNo, SeqNo) column pair. In other words, the query doesn't
have to check _every_ item line in order to determine the one with the
lowest sequence number.

Just a reminder for anyone else who happens to be reading - the SQL
statement that we are referring to has an incomplete JOIN. The correct
version was posted in a follow-up message to the list.

Regards,

John Taylor
Canada

+---
| This is the Midrange System Mailing List!
| To submit a new message, send your mail to MIDRANGE-L@midrange.com.
| To subscribe to this list send email to MIDRANGE-L-SUB@midrange.com.
| To unsubscribe from this list send email to MIDRANGE-L-UNSUB@midrange.com.
| Questions should be directed to the list owner/operator: david@midrange.com
+---