Hi Kevin,

the times seem very similar now.
But one thing I'ld always do in similar timing/benchmarking
is variate the numbers of rows actually searched
(and here possibly as well the typical search depth...)

This way you'll see differences in the efficiency of
the implemented algorithm, if any exist, and perhaps
get a warning if the approach doesn't scale well.

Have you done any research into the speed of the .select
of datatables or creation of DataViews?

As compared to a select against a local cursor or a set filter?

my 0.02 EUR

thomas