MySQL delete records between 'n' rows - mysql

Question

I am trying to select records between let's say 2nd and 5th row. My id's are not in sequentially order so I tried to retrieve the row_number this way:
SELECT #curRow := #curRow + 1 AS row_number
FROM product_image p
JOIN (SELECT #curRow := 0) r
WHERE row_number BETWEEN 2 AND 5
My table is kind of:
id name
23 A
42 B
98 C
102 D
109 E
What I get is row_number column doesn't exist and it really does not exist. But how can retrieve the records between 2 and 5 ( 3th and 4th )? I red similar post but didn't understand the query well. Thank you in advance!

Answer 1

The reason the query fails is that you can't use column aliases in a WHERE clause. You can however use them in a GROUP BY and HAVING:
SELECT *, #curRow := #curRow + 1 AS row_number
FROM product_image p
JOIN (SELECT #curRow := 0) r
GROUP BY row_number
HAVING row_number between 2 and 5

Answer 2

Since you don't really need to select the row number, there is no need to rely on undocumented "features" like #curRow := #curRow + 1. Just use LIMIT and OFFSET. To select the rows from 2 to 5 you would just need
select id
from product_image
order by name, id
limit 4
offset 1
Note that you need a well defined ORDER BY clause. Otherwise the result can depend on what index has been used by the engine.
To delete those rows, use a DELETE .. JOIN statement
delete product_image
from product_image
join (
select id
from product_image
order by name, id
limit 4
offset 1
)x using(id)
Demo: http://sqlfiddle.com/#!9/66fa4/1
No tricks. No complex queries.

Answer 3

You can do a DELETE with JOIN:
DELETE p
FROM product_image p
INNER JOIN (
SELECT id, #curRow := #curRow + 1 AS row_number
FROM product_image p
CROSS JOIN (SELECT #curRow := 0) r
ORDER BY id
) AS t ON p.id = t.id
WHERE t.row_number IN (2,3)
Derived tabe t is created using your query. You can join to this table in order to identify any records you want and delete them.
Demo here

Answer 4

Check this -
http://sqlfiddle.com/#!9/c462d/5
select * from
(
SELECT #curRow := #curRow + 1 AS row_number,p.*
FROM product_image p
JOIN (SELECT #curRow := 0) r)
a
WHERE a. row_number BETWEEN 2 AND 5

Answer 5

To add a synthetic row number to a query, you need a subquery. Like so.
SELECT id FROM (
SELECT p.*, #curRow := #curRow + 1 AS row_number
FROM product_image p
JOIN (SELECT #curRow := 0) r
) q
WHERE row_number BETWEEN 2 AND 5
Then, you can, if you wish, use the result to drive a delete operation.
DELETE FROM product_image
WHERE id IN (
SELECT id FROM (
SELECT p.*, #curRow := #curRow + 1 AS row_number
FROM product_image p
JOIN (SELECT #curRow := 0) r
) q
WHERE row_number BETWEEN 2 AND 5
)
But this is an extremely bad thing to do. Why? You're relying on a particular order in the result set of your inner query. The rows in a SQL result set are, without any ORDER BY clause, returned in an unpredictable order. Server optimizers exploit this. Unpredictable is like random, but worse. Random implies that it's unlikely the order will be the same each time you run the query. Random here is good because you'll catch your problem during testing. Unpredictable, on the other hand, means the order remains the same until it doesn't. If you're not sure why that's bad, look up Murphy's law.