Bruce used my spu_generateinsert script today and identified a further issue, it didn't handle GUIDs properly. I fixed that and did a spot of further testing on one of my test databases and discovered a collation issue, so I fixed that as well whilst I was at it. The result is that it runs slower now, but specifies COLLATE database_default on each column to ensure that there are no "Cannot resolve collation conflict for column 1 in SELECT statement" errors raised.

As before, download it here and feel free to feed back any issues/improvements etc.