Kevin,

There is, the main comparison being highlighted now appears to be how quickly the SQL will execute.

It is a shame we cannot reproduce this ourselves, as there are lots of possible variables apart from the code inefficiencies already discussed.

One is the disk spanning used by VFP. If VFP is resource constrained, it may span your resultset to disk unlike the C# example that will create a memory-resident resultset. That alone may explain the performance difference. Ironically it might even be the simultaneous presence of the C# version that forces VFP to span the resultset to disk. This may explain why you saw different results from one day to the next.

Such variables are why scientific proof requires a result to be reproducable by others. For example, I posted a reproducable example where using APPEND rather than SCATTER/INSERT made a huge difference with a 150-row, 2000-record table. With just one field and 1000 records, the same example yields a negligible difference as you report. That would be useful knowledge for people producing 1000-record resultsets with just a primary key ;-)

If your example could be reproduced in other infrastructures, it would form new, interesting knowledge. I'd really like to see it defined one way or the other- I want to know, as my experience with much larger resultsets was completely different from yours. As it is, those who leap upon your example as a "proof" are characterizing themselves.