Converting MySQL query that has a GROUP BY to MSSQL query - mysql

Question

Hi I'm trying to convert my MySQL database to MSSQL one so what I'm doing is changing most of my queries. Then I encountered the GROUP BY problem in the MSSQL so I was stuck in this part.
tbl_assign role:
tar_id int IDENTITY(1,1) PRIMARY KEY,
tar_auth int NOT NULL,
tar_role varchar(255) NOT NULL,
tar_group int NOT NULL,
tar_division int NOT NULL,
tar_add varchar(255) NOT NULL,
tar_edit varchar(255) NOT NULL,
tar_view varchar(255) NOT NULL,
tar_delete varchar(255) NOT NULL,
tar_updated_date datetime NOT NULL,
tar_updated_by int NOT NULL,
tar_owner_id int NOT NULL,
tar_assign_date datetime NOT NULL,
tar_is_deleted tinyint NOT NULL
tbl_user:
u_id int NOT NULL,
u_uname varchar(255) NOT NULL,
u_pword varchar(255) NOT NULL,
u_fname varchar(255) NOT NULL,
u_mname varchar(255) NOT NULL,
u_lname varchar(255) NOT NULL,
u_spark_id varchar(255) NOT NULL,
u_email varchar(255) NOT NULL,
u_mobile decimal(11,0) NOT NULL,
u_address varchar(255) NOT NULL,
u_div int NOT NULL,
u_group int NOT NULL,
u_role int NOT NULL,
u_updated_by int NOT NULL,
u_updated_date datetime NOT NULL,
u_reg_date datetime NOT NULL,
is_active tinyint NOT NULL,
is_online tinyint NOT NULL,
u_photo varchar(MAX) NOT NULL
Sample data
tbl_user:
u_id - 20
u_uname - USERNAME
u_pword - *********
u_fname - Fname
u_mname - Mname
u_lname - Lname
u_spark_id - 1
u_email - email#sample.com
u_mobile - 0
u_address - Address
u_div - 0
u_group - 0
u_role - 0
u_updated_by - 1
u_updated_date - 2015-07-30 00:00:00
u_reg_date - 2015-04-08 00:00:00
is_active - 1
is_online - 1
u_photo - 0
tbl_assign_role:
tar_id - 264
tar_auth - 2
tar_role - 44
tar_group - 1
tar_division - 1
tar_add - 1-1-1-1-1-1-1-0-0-1-0-1-1-0-0
tar_edit - 1-1-1-1-1-1-1-0-0-1-0-1-1-0-0
tar_view - 1-1-1-1-1-1-1-0-0-1-0-1-1-0-0
tar_delete - 0-0-0-0-0-0-0-0-0-0-0-0-0-0-0
tar_updated_date - 2015-04-08 11:51:50.000
tar_updated_by - 17
tar_owner_id - 20
tar_assign_date - 2015-04-08 11:51:50.000
tar_is_deleted - 0
The tbl_assign_role has 9 more entries that has the same owner_id but different in the other columns.
MySQL query:
SELECT * FROM tbl_user
JOIN tbl_assign_role ON tbl_user.u_id = tbl_assign_role.tar_owner_id
WHERE is_active = 1
AND u_id !=1
AND tar_is_deleted = 0
GROUP BY tbl_assign_role.tar_owner_id
ORDER BY tbl_user.u_updated_date DESC
Result of MySQL query has only one row. It will give the first entry on the tbl_assign_role that has the same owner_id.
MSSQL query:
SELECT * FROM tbl_user
JOIN tbl_assign_role ON tbl_user.u_id = tbl_assign_role.tar_owner_id
WHERE is_active = 1
AND u_id !=1
AND tar_is_deleted = 0
ORDER BY tbl_user.u_updated_date DESC
Result of MSSQL query has ten rows. It will give all the result of the tbl_assign_role that has the same owner_id.
How can I achieve the same result in the MySQL query?

Answer 1

For many databases, and MS SQL Server is one, it is not possible to specify just a single column in the GROUP BY clause, but use the wildcard for all columns in the select clause. MySQL is the odd one here, it does allow this.
So, the message is simple. You MUST specify the non-aggregating columns in the GROUP BY clause. No exceptions.
You might want to consider use of ROW_NUMBER() OVER([partition by ...] order by ...) if you are really just wanting to return a single row per tar_owner_id.
SELECT
*
FROM (
SELECT
*
, ROW_NUMBER() OVER (PARTITION BY tbl_assign_role.tar_owner_id
ORDER BY tbl_user.u_updated_date DESC) AS rn
FROM tbl_user
JOIN tbl_assign_role
ON tbl_user.u_id = tbl_assign_role.tar_owner_id
WHERE is_active = 1
AND u_id != 1
AND tar_is_deleted = 0
) AS D
WHERE D.rn = 1
ORDER BY <<some column(s)>>
OFFSET 10 ROWS
FETCH NEXT 10 ROWS ONLY