mysql select based on two criteria

Question

I'm a little rushed for time so I know I could look for the answer myself but stackoverflow is faster! :)

I have a table with data let's pretend it is product information. the table keeps track of when the product data is updated. If I run a query like

SELECT * FROM productChangeLog where change = 'price changed'

I get results like

id  product  date       change
---------------------------------------------    
236 product1 03/14/2011 'price changed'    
241 product2 03/14/2011 'price changed'    
242 product2 03/14/2011 'description changed'    
512 product1 05/16/2011 'price changed'    
517 product1 05/16/2011 'description changed'

what I want to do is select only the most recent 'price change' for each product. what do I need to add to my query so that i would only get row 241 and 512? the higher id is always the more recent change

thanks so much!

Answer 1

SELECT t.max_id, t.product, pcl.date, pcl.change
    FROM (SELECT product, MAX(id) AS max_id
              FROM productChangeLog 
              WHERE change = 'price changed'
              GROUP BY product) t
        INNER JOIN productChangeLog pcl
            ON t.max_id = pcl.id

Answer 2

GROUP BY the key-fields and use HAVING clause to extract the record.

In your case, you may try,

-- Don't use this if the product data can be updated twice on
-- the same day since the date field doesn't have a timestamp
SELECT * FROM productChangeLog where change = 'price changed' 
GROUP BY product HAVING MAX(date) = date

OR

-- Assuming the id is an integer field, this might be faster
SELECT * FROM productChangeLog where change = 'price changed' 
GROUP BY product HAVING MAX(id) = id

These are illegal queries in standard SQL, but should work in MySQL.

BTW, how is your result showing row 242 and 517 with 'description changed'?

Answer 3

This approach accomplishes the same thing, but uses ANSI syntax and can be extended to incorporate multiple fields:

SELECT
  *
FROM
  Table AS T1
WHERE
  NOT EXISTS
    (
    SELECT * FROM Table AS T2 WHERE T1.Group = T2.Group AND T2.Date < T1.Date
    )

If { Group, Date } is not unique, you need just a little more code:

SELECT
  *
FROM
  Table AS T1
WHERE
  NOT EXISTS
    (
    SELECT * FROM Table AS T2 WHERE T1.Group = T2.Group AND (T2.Date < T1.Date OR (T1.Date = T2.Date AND T2.ID < T1.ID))
    )