Develop Reference

Return last inserted ID without using a second query

I'm working on an ASP.NET project (C#) with SQL Server 2008.
When I insert a row into a table in the database, I would like to get the last inserted ID, which is the table's IDENTITY (Auto Incremented).
I do not wish to use another query, and do something like...
SELECT MAX(ID) FROM USERS;
Because - even though it's only one query - it feels lame...
When I insert something I usually use ExecuteNonQuery(), which returns the number of affected rows.
int y = Command.ExecuteNonQuery();
Isn't there a way to return the last inserted ID without using another query?

Most folks do this in the following way:
INSERT dbo.Users(Username)
VALUES('my new name');
SELECT NewID = SCOPE_IDENTITY();
(Or instead of a query, assigning that to a variable.)
So it's not really two queries against the table...
However there is also the following way:
INSERT dbo.Users(Username)
OUTPUT inserted.ID
VALUES('my new name');
You won't really be able to retrieve this with ExecuteNonQuery, though.

You can return the id as an output parameter from the stored procedure, e.g. #userId int output
Then, after the insert, SET #userId = scope_identity()

even though it's only one query - it feels lame...
It actually is also wrong as you can have multiple overlapping iserts.
That is one thing that I always fuind funny - people not reading the documentation.
SELECT SCOPE_IDENTITY()
returns the last identity value generated in a specific scope and is syntactically correct. It also is properly documented.
Isn't there a way to return the last inserted ID without using another query?
Yes. Ask for the number in the saame SQL batch.
INSERT (blablab9a); SELECT SCOPE_IDENTITY ();
as ONE string. ExecuteScalar.
You can have more than one SQL statement in one batch.

If you want to execute query from C# code & want to get last inserted id then you have to find the following code.
SqlConnection connection = new SqlConnection(System.Configuration.ConfigurationManager.ConnectionStrings["ConnectionString"].ConnectionString);
connection.Open();
string sql = "Insert into [Order] (customer_id) values (" + Session["Customer_id"] + "); SELECT SCOPE_IDENTITY()";
SqlCommand cmd = new SqlCommand();
cmd.Connection = connection;
cmd.CommandText = sql;
cmd.CommandType = CommandType.Text;
var order_id = cmd.ExecuteScalar();
connection.Close();
Console.Write(order_id);

ExecuteScalar returns value is not suitable

I use Enterprise Library and I have one problem:
string sql = "
UPDATE StackOverflow SET UserName = #UserName
WHERE Id = #Id
";
DbCommand cmd = base.Database.GetSqlStringCommand(sql);
base.Database.AddInParameter(cmd, "Id", DbType.Int32, StackOverflow.Id);
base.Database.AddInParameter(cmd, "UserName", DbType.Int32, StackOverflow.UserName);
int val = Convert.ToInt32(base.Database.ExecuteScalar(cmd));
Convert.ToInt32(base.Database.ExecuteScalar(cmd)) //returns 0.
I've read this article http://msdn.microsoft.com/en-us/library/system.data.sqlclient.sqlcommand.executescalar.aspx
The article says:
The function returns the new Identity column value if a new row was inserted, 0 on failure.
but I did not insert into that table - I only want to update and return updated row Id.

You should use ExecuteNonQuery in your case.
ExecuteScalar
Executes the query, and returns the first column of the first row in the result set returned by the query
ExecuteNonQuery
Executes a Transact-SQL statement against the connection and returns the number of rows affected
Your query doesn't return anything, so ExecuteScalar is not the right method to work with.
ExecuteNonQuery on the other side will give the correct information if your query has updated anything.

If you modify your SQL statement to the following, I think this will give you the result your expecting:
string sql = "
UPDATE StackOverflow SET UserName = #UserName
WHERE Id = #Id
RETURN #Id
";

C# - SELECT Query number of rows affected is always -1

I have this code and it always returns -1.I have three tables (a picture is more suggestive ):
I want to see if the row is already in the ReservationDetails table, and if it's not to insert it.
try
{
SqlConnection conn = new SqlConnection...
SqlCommand slct = new SqlCommand("SELECT * FROM ReservationDetails WHERE rID=#rID AND RNumber=#RNumber", conn);
slct.Parameters.AddWithValue("#rID", (int)comboBox1.SelectedValue);
slct.Parameters.AddWithValue("#RNumber", dataGridView1.SelectedRows[0].Cells[0].Value);
int noRows;//counts if we already have the entry in the table
conn.Open();
noRows = slct.ExecuteNonQuery();
conn.Close();
MessageBox.Show("The result of select="+noRows);
if (noRows ==0) //we can insert the new row

Have you read the documentation of SqlCommand.ExecuteNonQuery?
For UPDATE, INSERT, and DELETE statements, the return value is the number of rows affected by the command. When a trigger exists on a table being inserted or updated, the return value includes the number of rows affected by both the insert or update operation and the number of rows affected by the trigger or triggers. For all other types of statements, the return value is -1. If a rollback occurs, the return value is also -1.
And your query is SELECT.

You should
1) Change your TSQL to
SELECT COUNT(*) FROM ReservationDetails WHERE ...
(better still, use IF EXISTS ...)
2) and use ExecuteScalar():
noRows = (int) slct.ExecuteScalar();
Also: you will need to use a transaction (or some other atomic technique), or else someone could insert a row in-between you testing and trying to insert it...
All that said, it would be better to create a stored procedure that given your parameters, atomically tests and inserts into the table, returning 1 if successful, or 0 if the row already existed.

It is better to do it in a single query so that you do not need to request server two times.
Create a procedure like this and call it from the code.
IF EXISTS (SELECT 1 from ReservationDetails WHERE rID=#rID AND RNumber=#RNumber)
BEGIN
insert into ReservationDetails values(#rID,#RNumber)
END

As per Microsoft:
You can use the ExecuteNonQuery to perform catalog operations (for example, querying the structure of a database or creating database objects such as tables), or to change the data in a database without using a DataSet by executing UPDATE, INSERT, or DELETE statements.
What you may need, instead of ExecuteNonQuery is ExecuteScalar and put the COUNT in your select query.
i.e.
SqlCommand slct = new SqlCommand("SELECT COUNT(*) FROM ReservationDetails WHERE rID=#rID AND RNumber=#RNumber", conn);
Also, try to make use of the using statement in C#, so you don't need to worry about closing the connection manually, even if things fail.
i.e.
using (SqlConnection conn = new SqlConnection(connString))
{
SqlCommand cmd = new SqlCommand(sql, conn);
try
{
conn.Open();
newProdID = (Int32)cmd.ExecuteScalar();
}
catch (Exception ex)
{
//Do stuff
}
}
see:
http://msdn.microsoft.com/en-us/library/system.data.sqlclient.sqlcommand.executescalar.aspx

#nickNatra
When ever you use
Select command
It will return you values. Which can be either used by
DataSet or SqlDataReader
But
command.ExecuteNonQuery()
is used only when you are using
Insert , Update , Delete where the Rows are getting effected in your table
Yes, If you do want to know how much records are there in your query.
You can perform
a) Modify your query "select count(*) from table"
where you will only get one value ie. Number of Rows.
b) Using this query perform command. ExecuteScalar() which will return only First row and first column which is the Row Count
Hence this satisfy's your requirement.
Cheers!!

Retrieve serial ID with Npgsql when inserting with ExecuteScalar

I'm trying to insert a row into a PostgreSQL table with a serial primary key and I need to retrieve this column after it was inserted. I got something like this:
The table "pais" has 3 columns: id, pais, capital; id is a serial column and is its primary key.
NpgsqlCommand query = new NpgsqlCommand("insert into pais(nombre, capital) values(#nombre, #capital)", conn);
query.Parameters.Add(new NpgsqlParameter("nombre", NpgsqlDbType.Varchar));
query.Parameters.Add(new NpgsqlParameter("capital", NpgsqlDbType.Varchar));
query.Prepare();
query.Parameters[0].Value = this.textBox1.Text;
query.Parameters[1].Value = this.textBox2.Text;
Object res = query.ExecuteScalar();
Console.WriteLine(res);
It inserts the row on the table but "res" value is null. If I insert with the nexval('table_sequence') also returns null.
Any idea of how can I return the id of the table? Am I missing something?
Thanks in advance

Is that thread safe?
What if another insert happens between your insert and select?
Why not use:
INSERT INTO table (fieldnames) VALUES (values) RETURNING idcolumn?

insert into pais(nombre, capital) values(#nombre, #capital) RETURNING id
replace id with your primary keyenter code here and use
Object res = query.ExecuteScalar();
Inside res you'll have the PK.

In order to select the last identity inserted you need to use: currval(sequencename)
so your select statement should look like:
NpgsqlCommand query = new NpgsqlCommand("insert into pais(nombre, capital) values(#nombre, #capital);select currval('table_sequence');", conn);

The insert itself does not cause a value to be returned. When you perform ExecuteScalar it is looking for a single value to be "Selected" so to speak.
I believe you need to follow up your insert with a select statement to solve your issue.
If you were using t-sql you would do this like so
string sql =
"INSERT INTO [Table] (FieldName) VALUES (#ParamName); "
+ "SELECT CAST(scope_identity() AS int)";
ExecuteScalar would then return the unique id;
I am not sure of the exact syntax for postGresql but hopefully this allows you to solve your issue.

How to get the generated id from an inserted row using ExecuteScalar?

I know that in Oracle I can get the generated id (or any other column) from an inserted row as an output parameter.
Ex:
insert into foo values('foo','bar') returning id into :myOutputParameter
Is there a way to do the same, but using ExecuteScalar instead of ExecuteNonQuery?
I don't want to use output parameters or stored procedures.
ps: I'm using Oracle, not sql server!!!

If you are on oracle, you have to use ExecuteNonQuery and ResultParameter. There is no way to write this as query.
using (OracleCommand cmd = con.CreateCommand()) {
cmd.CommandText = "insert into foo values('foo','bar') returning id into :myOutputParameter";
cmd.Parameters.Add(new OracleParameter("myOutputParameter", OracleDbType.Decimal), ParameterDirection.ReturnValue);
cmd.ExecuteNonQuery(); // an INSERT is always a Non Query
return Convert.ToDecimal(cmd.Parameters["myOutputParameter"].Value);
}

Oracle uses sequences as for his identity columns, if we may say so.
If you have set a sequence for your table primary key, you also have to write a trigger that will insert the Sequence.NextValue or so into your primary key field.
Assuming that you are already familiar with this concept, simply query your sequence, then you will get your answer. What is very practiced in Oracle is to make yourself a function which will return an INT, then within your function, you perform your INSERT. Assuming that you have setup your trigger correctly, you will then be able to return the value of your sequence by querying it.
Here's an instance:
CREATE TABLE my_table (
id_my_table INT PRIMARY KEY
description VARCHAR2(100) NOT NULL
)
CREATE SEQUENCE my_table_seq
MINVALUE 1
MAXVALUE 1000
START WITH 1
INCREMENT BY 2
CACHE 5;
If you want to manage the auto-increment yourself, here's how:
INSERT INTO my_table (
id_my_table,
description
) VALUES (my_table_seq.NEXTVAL, "Some description");
COMMIT;
On the other hand, if you wish not to care about the PRIMARY KEY increment, you may proceed with a trigger.
CREATE OR REPLACE TRIGGER my_table_insert_trg
BEFORE INSERT ON my_table FOR EACH ROW
BEGIN
SELECT my_table_seq.NEXTVAL INTO :NEW.id_my_table FROM DUAL;
END;
Then, when you're inserting, you simply type the INSERT statement as follows:
INSERT INTO my_table (description) VALUES ("Some other description");
COMMIT;
After an INSERT, I guess you'll want to
SELECT my_table_seq.CURRVAL
or something like this to select the actual value of your sequence.
Here are some links to help:
http://www.orafaq.com/wiki/Sequence
http://www.orafaq.com/wiki/AutoNumber_and_Identity_columns
Hope this helps!

You can use below code.
using (OracleCommand cmd = conn.CreateCommand())
{
cmd.CommandText = #"INSERT INTO my_table(name, address)
VALUES ('Girish','Gurgaon India')
RETURNING my_id INTO :my_id_param";
OracleParameter outputParameter = new OracleParameter("my_id_param", OracleDbType.Decimal);
outputParameter.Direction = ParameterDirection.Output;
cmd.Parameters.Add(outputParameter);
cmd.ExecuteNonQuery();
return Convert.ToDecimal(outputParameter.Value);
}

one possible way if one can add one column named "guid" to the table :
when inserting one record from c#, generate a guid and write it to the guid column.
then perform a select with the generated guid, and you have got the id of inserted record :)

Select t.userid_pk From Crm_User_Info T
Where T.Rowid = (select max(t.rowid) from crm_user_info t)
this will return your required id