Thanks for the feedback, Hugo,
I explained why a merry-go-round scan is not possible for this in the article. I even demonstrated what a merry-go-round scan is... updates don't do merry-go-round scans on Clustered indexes especially when the index "hint" forces the index to be use. I put "hint" in parenthesis because when it comes to index hints, it's not a hint... it's a mandatory directive. MS even warns against such a practice for "normal" queries because it overrides the optimizer and most people aren't smarter than the optimizer.
I also appreciate your extreme testing and reporting back on you found. That's awesome. Thanks.
If anyone can cause the code to fail on an unpartitioned table, I'd sure be interested in seeing it... I'm not beyond publishing a retraction of my findings if I'm proven wrong or publishing an update to the article for an additional "exception."
Thanks folks.
--Jeff Moden
Change is inevitable... Change for the better is not.