I like both approaches to the problem. They both have a refreshing spark of originality.

I'm not entirely sure if I completely agree with all Hugo's objections, though I share his caution. 'Quirky Update' techniques can go wrong if you are not entirely aware of certain 'gotchas', but they were documented, used, and supported even before Microsoft bought the product. For the 'quirky update' approach that Lynne uses to work safely, you have to remember certain rules (e.g. updates are done in the order of the clustered index, all '@var=col=expression' variable assignments are done before any simple column updates-I hope I've remembered that the right way around!)