Kevin, it's just a trivial example.

Suffice to say that some of us build cursors of indeterminate size and hammer them hard. We tried it as a SP but to get good performance without annoying all other users you'd have to put the SP in its own database on a much larger machine before you do anything else. You get to the point when you have to ask why you are going to all this time and trouble when there is a viable, reliable, tested existing mechanism that runs at high speed on the PC/s where it is needed and without needing 99% of the features offered by SQL Server, apart from its ability to handle resultsets of indeterminate size. ;-) Apart from the "SP is always best" mantra with which I disagreed (as you know) long before it became fashionable to do so ;-) there is no rational reason to spend big $ and take risk fixing something that isn't broken.

In view of previous discussion on this topic and agreement that disk spanning is a key distinguishing area to which attention would be paid in Linq, maybe it would have been better if Linq was delivered as a migration option before the announcement was made. Such is life.