SQL - Inserting a row and returning primary key

Question

I have a little witty problem. Say, I inserted a row with some data in a table where a primary key is present. How would one \"SELECT\" the primary key of the row one just i

Answer 1

select MAX(id_column) from table

That, in theory, should return you that last inserted id. If it's a busy database with many inserts going on it may not get the one you just did but another.

Anyhow, an alternative to other methods.

Answer 2

If you need to retrieve the new index in MS SQL when there are triggers on the table then you have to use a little workaround. A simple OUTPUT will not work. You have to do something like this (in VB.NET):

DECLARE @newKeyTbl TABLE (newKey INT);
INSERT INTO myDbName(myFieldName) OUTPUT INSERTED.myKeyName INTO @newKeyTbl VALUES('myValue'); " & _
SELECT newKey FROM @newKeyTbl;"

If using .NET, then the return value from this query can be directly cast to an integer (you have to call "ExecuteScalar" on the .NET SqlCommand to get the return).

Answer 3

For SQL Server 2005 and up, and regardless of what type your primary key is, you could always use the OUTPUT clause to return the values inserted:

INSERT INTO dbo.YourTable(col1, col2, ...., colN)
OUTPUT Inserted.PrimaryKey
VALUES(val1, val2, ....., valN)

Answer 4

For Postgresql:

SELECT CURRVAL(pg_get_serial_sequence('schema.table','id'))

Source: https://stackoverflow.com/questions/2944297/postgresql-function-for-last-inserted-id

Answer 5

For SQLite:

SELECT [Column_1],  [Column_2],...  [Column_n]
FROM [YourTable]
WHERE rowid = (SELECT last_insert_rowid())

whereas:

• Column_1, Column_2,... Column_n: are the primary key of YourTable.

If you'd created YourTable with primary key replaced rowid (i.e. one column pk defined as INTEGER PRIMARY KEY) you just use:

SELECT last_insert_rowid()

Which is a common case.
Finally, this wont work for WITHOUT_ROWID tables.

Please Check:

https://www.sqlite.org/lang_corefunc.html#last_insert_rowid

Answer 6

For MySQL, use LAST_INSERT_ID()

http://dev.mysql.com/doc/refman/5.0/en/getting-unique-id.html

You should also be able to start a transaction, insert the row, and select the row using some field that has a unique value that you just inserted, like a timestamp or guid. This should work in pretty much any RDBMS that supports transactions, as long as you have a good unique field to select the row with.