> One of the major differences between .Net and VFP is that even when working against SQL Server the "cursor" in VFP is held in a temporary
> dbf (which for large datasets we actually point toward an imdisk ramdisk: nothing is faster than memory). In .Net, a recordset is help in memory.
> That's fine for small datasets, but datasets for reporting (munged from what's available on the backend server) can be quite large.
> It will be interested to see speed differences between the two approaches.

I have done an import process from Oracle and populate the result set inside a local Access database for some research organization using X#. Was 35million+ records spread over 7 different tables with relationships. It takes approximately 15 minutes to complete the whole process or just under 700 rows/second.

I have started also an InMemory SQL database (still in development), but benchmarks tests so far indicate processing of about 3.5million rows per second. The same queries for most SQL engines for the same queries flattens out at approx 28,000 rows per seconds. So in conclusion, I don't think speed is a concern for a cursor in X#.