SQL SELECT with GROUP BY: Retrieve records for a list of customers

Question

I need to write a SQL statement to retrieve a list of customers, which are in two tables of different natures, say T1 and T2.
The customer is a FK in both tables. Both tables have a column which is the date of an order.
INPUT:
* list of customers, say '12345, 34567, 45678'
OUTPUT
* REF CURSOR to a recordset containing: {custID, startDate, endDate}
Algorithm
* For each customer, find the date of first order (from T1 and T2)
* For that date, t0, startDate=t0-10 days; endDate := t0+15 days;
I have a feeling some GROUP BY is involved to do the trick.
Can it be done without any looping? I am just not experienced enough to solve this in an optimal way.
Any help, please?
Thank you.

Answer 1

Hi,
I'm not sure what your data is, or shat you want for output.
If your tables can have many rows for the same customer_id, but your output should have only one row per customer_id, then it sounds like you do need to "GROUP BY custiomer_id".
If the data is coming from two different tables, first do a UNION of those two tables in a sub-query (like union_data below), and then do a GROUP BY on the results of the sub-query.
It will probably be more efficient if you GROUP BY in the sub-query, too.
I believe you need something like this:
WITH     union_data     AS
     SELECT       customer_id
     ,       MIN (order_date)     AS start_dt
     FROM       t1
     WHERE       customer_id     IN (12345, 34567, 45678)
     GROUP BY  customer_id
UNION ALL
     SELECT       customer_id                    -- column names may be different from t1
     ,       MIN (order_date)     AS start_dt
     FROM       t2
     WHERE       customer_id     IN (12345, 34567, 45678)
     GROUP BY  customer_id
SELECT       customer_id
,       MIN (start_date)      AS start_date
,       MIN (start_date) - 10      AS t0
,       MIN (start_date) + 5      AS end_date
FROM       union_data
GROUP BY  customer_id;If you need help, post a little sample data (CREATE TABLE and INSERT statements for both tables, relevant columns only), and the results you want to get from that data.