Question

I know I can update a single record like this – but then how to I get access to the id of the record that was updated? (I’m using MSSQL so I can’t use Oracles RowId)

update myTable set myCol = 'foo' where itemId in (select top 1 itemId from myTable ) 

If I was peforming an Insert I could use getGeneratedKeys to get the id field value, but I don’t think there is an equivalent for an update?

I know I can use a scrollable resultset to do what I want

i.e.

stmt = conn.prepareStatement('select top 1 myCol, itemId from myTable', ResultSet.TYPE_SCROLL_SENSITIVE, ResultSet.CONCUR_UPDATABLE); ResultSet resultSet = stmt.executeQuery(); if(resultSet.first()){     resultSet.updateString(1, 'foo');     resultSet.updateRow();     String theItemId = resultSet.getString(1) } resultSet.close(); 

but I’m concerned about performance as testing shows lock timeouts under load and I was wondering if there was a better/simpler way?

— EDIT: Just to finalise this issue… When we migrate to MSSQL2005 we will upgrade our code to use Rich’s answer. In the current release we have used the lock hints: (UPDLOCK ROWLOCK READPAST) to mitigate the performance problems our original code showed.

Answer 1

This example works really well in MSSQL 2005…

SET ANSI_NULLS ON GO SET QUOTED_IDENTIFIER ON GO  DROP TABLE [dbo].[TEST_TABLE] GO  CREATE TABLE [dbo].[TEST_TABLE](     [id] [int] IDENTITY(1,1) NOT NULL,     [name] [nvarchar](100) COLLATE SQL_Latin1_General_CP1_CI_AS NOT NULL,  CONSTRAINT [PK_TEST_TABLE] PRIMARY KEY CLUSTERED  (     [id] ASC )WITH (PAD_INDEX  = OFF, STATISTICS_NORECOMPUTE  = OFF, IGNORE_DUP_KEY = OFF, ALLOW_ROW_LOCKS  = ON, ALLOW_PAGE_LOCKS  = ON) ON [PRIMARY] ) ON [PRIMARY]   -- An insert which will return the identity INSERT INTO [dbo].[TEST_TABLE] ([name])  OUTPUT inserted.id VALUES('Test 1')  -- Another insert which will return the identity INSERT INTO [dbo].[TEST_TABLE] ([name])  OUTPUT inserted.id VALUES('Test 2')  -- Now an update which will return the identity UPDATE [dbo].[TEST_TABLE] SET [name] = 'Updated Test 1' OUTPUT inserted.id WHERE [name] = 'Test 1'  SELECT id, [name] FROM [dbo].[TEST_TABLE] 

And more specifically to your query…

update myTable set myCol = 'foo' output inserted.itemid where itemId in (select top 1 itemId from myTable )