http://www.sqlskills.com/blogs/paul/sql-server-2008-new-undocumented-physical-row-locator-function/

Not so sure how happy I would be using this - but then as Paul says all the cool stuff is undocumented.

Great question, I really like how you have included the ability to retain the latest version.. I usually use a more simple way to remove duplicate rows..

Using the same table...

Declare @FName varchar(30), -- values to look for

@LName varchar (30),-- values to look for

@cnt int -- count

Declare getallrecords cursor local static For

Select count (1), Fname, Lname

from DuplicateRow (nolock)

group by FName, LName having count(1)>1

Open getallrecords

Fetch next from getallrecords into @cnt,@FName,@LName

--Cursor to check with all other records

While @@fetch_status=0

Begin

Set @cnt= @cnt-1

Set rowcount @cnt

-- Deleting the duplicate records. Observe that all fields are mentioned at the where condition

Delete from DuplicateRow where Fname=@FName and LName=@LName

Set rowcount 0

Fetch next from getallrecords into @cnt,@FName,@LName

End

Close getallrecords

Deallocate getallrecords

Obviously depending on how many rows you need to compare to ensure your only removing duplicates would depend on what you declare and select.