SQL how to get rows with max value of a column in the table in grouped rows

Question

I have a table with columns ('name', 'maj', 'min', 'patch') for each name there can be several major/minor/patch versions. For each name I need to get the major + minor values for the max 'patch' for that major/minor.

eg table 'data'

|id|name|maj|min|patch|
|1 |n1  |1  |0  | 1   |
|2 |n1  |1  |0  | 2   |
|3 |n1  |1  |0  | 3   |
|4 |n2  |1  |0  | 1   |
|5 |n2  |2  |0  | 1   |
|6 |n2  |2  |0  | 2   |

Expected output:

|id|name|maj|min|patch|
|3 |n1  |1  |0  | 3   |
|4 |n2  |1  |0  | 1   |
|6 |n2  |2  |0  | 2   |

I have only managed to get this data grouped by these fields and ordered but I don't know to get only the rows with max patch for each major/minor

SELECT id, name, maj, min, patch FROM data group by name, maj, min order by name, patch desc

Answer 1

The following query gets close to what you want:

SELECT
    name,
    maj,
    `min`,
    MAX(patch) AS patch
FROM data
GROUP BY
    name,
    maj,
    `min`;

But, this doesn't include the id column, which is not part of the aggregation query. If you want a general approach to include the entire matching rows, then consider using ROW_NUMBER :

WITH cte AS (
    SELECT *, ROW_NUMBER() OVER (PARTITION BY name, maj, `min` ORDER BY patch DESC) rn
    FROM data
)

SELECT id, name, maj, `min`, patch
FROM cte
WHERE rn = 1;

Answer 2

With NOT EXISTS:

select t.* from tablename t
where not exists (
  select 1 from tablename
  where name = t.name and maj = t.maj and min = t.min and patch > t.patch
)  

See the demo .
Results:

| id  | name | maj | min | patch |
| --- | ---- | --- | --- | ----- |
| 3   | n1   | 1   | 0   | 3     |
| 4   | n2   | 1   | 0   | 1     |
| 6   | n2   | 2   | 0   | 2     |

Answer 3

You can also do this with a correlated subquery:

selet t.*
from t
where t.patch = (select max(t2.patch)
                 from t t2
                 where t2.name = t.name and
                       t2.maj = t.maj and
                       t2.min = t.min
                );

If you have a large database, then an index on (name, maj, min, patch) will provide good performance.

Answer 4

You can simply using having Clause to achieve it Simple sample

SELECT id, name, maj, min, max(patch) AS max_patch FROM data GROUP BY patch DESC HAVING MAX(patch) >1;

Hope that's help you