Insert into inside cross apply - sql

Question

SELECT DISTINCT CODE
FROM T1
CROSS APPLY
(
INSERT INTO T4(TEXT1, TEXT2, TEXT3)
SELECT T2.TEXT1, T2.TEXT2, T3.TEXT3
FROM T2,
T3
WHERE T2.ID = T3.ID
AND T2.CODE = T1.CODE
) AS T
When using executing this query I get this error:
A nested INSERT, UPDATE, DELETE, or MERGE statement must have an OUTPUT clause.
What am I doing wrong?
EDIT
What I was planning to achieve was simulate a WHILE..LOOP.
Looping through all the CODEs in T1 and for each CODE get the TEXT fields from T2 and T3 (joining them with the ID) and insert them into table T4.
I am trying to separate the inserts by CODE because both tables have a large amount of data and I was trying to improve performance (maybe?!)

Answer 1

Kind of a guess
INSERT INTO FINAL (TEXT1 , TEXT2 , TEXT3)
SELECT DISTINCT T2.TEXT1, T2.TEXT2, T3.TEXT3
FROM T1
JOIN T2
on T2.ID = T1.ID
JOIN T3
on T3.VALUE = T2.VALUE

Answer 2

You can't do what you're attempting to do. The language will not allow you. To set the stage, I created the following 3 tables
SET NOCOUNT ON;
CREATE TABLE dbo.Final
(
ID int IDENTITY(1,1) NOT NULL
, text1 varchar(50) NOT NULL
, text2 varchar(50) NOT NULL
, text3 varchar(50) NOT NULL
);
CREATE TABLE dbo.Chain
(
ID int NOT NULL
);
CREATE TABLE dbo.Chained
(
ID int NOT NULL
, Foo int NOT NULL
);
Just to demo where the OUTPUT clause goes, we will insert 4 rows and see the nice pretty INSERTED virtual table and the associated ID values.
-- Works
INSERT INTO
dbo.Final
(
text1
, text2
, text3
)
OUTPUT
INSERTED.*
SELECT
D.t1
, D.t2
, D.t3
FROM
(
VALUES
('A', 'B', 'C')
, ('D', 'B', 'C')
, ('G', 'B', 'C')
, ('J', 'B', 'C')
) D (t1,t2,t3);
Now, if I perform the following statement, it will work fine. However, if I neglect the INSERT just to visually inspect what I want to put into the table, SQL Server will raise the following error
A nested INSERT, UPDATE, DELETE, or MERGE statement is not allowed in a SELECT statement that is not the immediate source of rows for an INSERT statement.
-- Comment out the insert portion to generate
-- A nested INSERT, UPDATE, DELETE, or MERGE statement is not allowed in a SELECT statement that is not the immediate source of rows for an INSERT statement.
INSERT INTO
dbo.Chain
(
ID
)
SELECT
X.ID
FROM
(
INSERT INTO
dbo.Final
(
text1
, text2
, text3
)
OUTPUT
INSERTED.*
SELECT
D.t1
, D.t2
, D.t3
FROM
(
VALUES
('A', 'B', 'C')
, ('D', 'B', 'C')
, ('G', 'B', 'C')
, ('J', 'B', 'C')
) D (t1,t2,t3)
) x
But, you want to go the extra mile and APPLY, or JOIN the results of that virtual table with something else and that's not going to fly.
A nested INSERT, UPDATE, DELETE, or MERGE statement is not allowed on either side of a JOIN or APPLY operator.
I suppose it's just one level of complexity too much.
-- Now, try the same bit except we use the derived table as a JOIN/APPLY
-- Can't fix what's not supported
-- A nested INSERT, UPDATE, DELETE, or MERGE statement is not allowed on either side of a JOIN or APPLY operator.
INSERT INTO
dbo.Chained
(
ID
, Foo
)
SELECT
X.ID
, D.foo
FROM
(
VALUES
(1)
) D(Foo)
CROSS APPLY
(
INSERT INTO
dbo.Final
(
text1
, text2
, text3
)
OUTPUT
INSERTED.*
SELECT
D.t1
, D.t2
, D.t3
FROM
(
VALUES
('A', 'B', 'C')
, ('D', 'B', 'C')
, ('G', 'B', 'C')
, ('J', 'B', 'C')
) D (t1,t2,t3)
) x;
If you really need something like that, then you'll have to break it out into separate statements.