Question

My table is called items. It has the columns item_id and source as well as some other columns.

I want to do a select ordered by item_id but I also want to not have any duplicates in the source column of the result.

What’s wrong with this query?

SELECT *
FROM items
WHERE item_section='sp_500'
ORDER BY item_id DESC
GROUP BY source
LIMIT 3

Answer 1

Should work like this:

SELECT *
FROM   items
JOIN  (
    SELECT max(item_id) AS item_id
    FROM   items
    WHERE  item_section = 'sp_500'
    GROUP  BY source
    ORDER  BY 1 DESC
    LIMIT  3
    ) i USING (item_id)

Why?

1) Filter by item_section = 'sp_500'

2) Collapse multiple items with the same source in a GROUP BY, because:

not have any duplicates in the source column

I take the biggest item_id per source – seems most plausible and you did not specify.

3) ORDER BY item_id DESC to get the greatest ones and LIMIT 3 (without dupes by now).

4) JOIN to the original table to get the whole row for the selected item_ids.