I tried to write a stored procedure that first inserts a new record into table and then returned the id of this new record. I am not sure if it is the correct way and the best w
To return a single scalar value to the caller you should use an OUTPUT
parameter, not RETURN
. RETURN
is for error/status codes. Also the prefix sp is redundant and unnecessary.
CREATE PROCEDURE dbo.AddAsset
@Name VARCHAR(500),
@URL VARCHAR(2000),
@new_identity INT = NULL OUTPUT
AS
BEGIN
SET NOCOUNT ON;
INSERT dbo.Assets(Name, URL) SELECT @Name, @URL;
SET @new_identity = SCOPE_IDENTITY();
END
GO
Then to call it:
DECLARE @new_identity INT;
EXEC dbo.AddAsset @Name = 'a', @URL = 'b', @new_identity = @new_identity OUTPUT;
PRINT @new_identity;
EDIT just adding a disclaimer that won't affect the asker in this specific scenario, but may help in other scenarios or for future readers. In SQL Server 2008 R2 and earlier, there is a potentially nasty bug with built-in functions such as SCOPE_IDENTITY
when parallelism is used to derive the results to be inserted (think INSERT FROM othertable
). This bug (here is the Connect item) is fixed in Cumulative Update #5 for SQL Server 2008 R2 SP1, but so far a fix has not appeared for 2008 R2 RTM, 2008 or 2005.