Yet Another Deleting Duplicate Records (in-place)

Here is another "deleting duplicate records" script, this time in-place. It exploits the ability to relatively easily add and drop a uniqueidentifier column to a table. The term "must-be-unique columns" refers to the set of columns that have to have a unique combination of values (natural primary key candidate for the table). The DELETE statement is quite generic and the comments suggest the changes needed for any other table. In addition, you obviously have to replace "MyTab" with the name of your table.

Note: The "DELETE" statement is designed and formatted in order to extend its usage - replacing "DELETE t" with "SELECT t.*" will return the records that will be deleted.